A client that stops sending after the app answers is timed out (closes #130)

When the app answered before the client had sent its whole body, and its
answer ended with its last bytes, every timeout was stopped,
SWWAF_CLIENT_REQUEST_TIMEOUT included. The transport went on waiting to
read the body, and Go's server cannot send the answer until that read
returns, so a client that stopped sending held the request until it
closed the connection. The timeouts now run on until the client has sent
its body, as they do before the app answers: once
SWWAF_CLIENT_REQUEST_TIMEOUT runs out the client is sent the app's
answer, the connection is closed, and the log line says timed_out.

Model: opus-5-5

Dockerfile

@@ -36,11 +36,12 @@ RUN go mod tidy -diff || \
     { echo "go.mod or go.sum is not tidy: run make tidy" >&2; exit 1; }
 
 # Go's build cache is kept on a tmpfs, out of the image: nothing uses it
-# after this step, and writing it into the image takes seconds.
+# after this step, and writing it into the image takes seconds. The tests
+# are built with the no_fs_access tag, as the binary is in the build stage.
 RUN --mount=type=tmpfs,target=/root/.cache/go-build \
-    go test -timeout 90s -race -cover ./... || \
+    go test -tags no_fs_access -timeout 90s -race -cover ./... || \
     { echo "--- Rerunning with -v for details ---"; \
-      go test -timeout 90s -race -v ./...; exit 1; }
+      go test -tags no_fs_access -timeout 90s -race -v ./...; exit 1; }
 
 # Tidy stage: `go mod tidy` in the test phase's Go, so that the files it
 # writes pass the test phase's check. Nothing else depends on it, so only
@@ -84,7 +85,10 @@ COPY . .
 # The VERSION build arg when one is given, otherwise
 # `git describe --tags --always` on the .git in the build context. With
 # .git present, a version that is still empty, dev or unknown fails the
-# build: git is missing or could not read the checkout.
+# build: git is missing or could not read the checkout. The no_fs_access
+# tag keeps Coraza from writing the files of a multipart body to the
+# system's temporary directory, since smallwebwaf writes only to its state
+# directory.
 ARG VERSION
 RUN VERSION="${VERSION:-$(git describe --tags --always)}"; \
     if [ -e .git ]; then \
@@ -93,7 +97,7 @@ RUN VERSION="${VERSION:-$(git describe --tags --always)}"; \
             exit 1 ;; \
         esac; \
     fi; \
-    CGO_ENABLED=0 go build -trimpath \
+    CGO_ENABLED=0 go build -tags no_fs_access -trimpath \
         -ldflags="-s -w -X main.Version=${VERSION}" \
         -o /usr/local/bin/smallwebwaf ./cmd/smallwebwaf
 

README.md

@@ -64,11 +64,10 @@ cannot read, serves Prometheus metrics to a scraper that holds the metrics
 token, lets an admin who holds the admin token list, add and lift bans and ask
 what it knows of a client, and in `observe` mode passes on the requests it would
 refuse, logging what it would have done with them. It comes as the image the
-app's own image is built on. The Core Rule Set reads no request body yet:
+app's own image is built on. The Core Rule Set reads a request's body too once
-`SWWAF_WAF_BODY_LIMIT`, which switches that on, comes with
+`SWWAF_WAF_BODY_LIMIT` is set. The rest of the design comes next, in the order
-https://git.eeqj.de/sneak/smallwebwaf/issues/116. The rest of the design comes
+of the build order in [`SPEC.md`](SPEC.md). The survey of existing tools that
-after that, in the order of the build order in [`SPEC.md`](SPEC.md). The survey
+led to the design is in [`EVALUATION.md`](EVALUATION.md).
-of existing tools that led to the design is in [`EVALUATION.md`](EVALUATION.md).
 
 ## Getting started
 
@@ -98,7 +97,9 @@ in `bin/state` unless `SWWAF_STATE_DIR` is set, and the default rule file of
 
 - Passes each request to the app and the app's answer back unchanged: method,
   path, query, headers, body and status. Bodies stream through in both
-  directions and are never held whole in memory. A WebSocket, or any other
+  directions and are never held whole in memory, but for the part of a request
+  body the Core Rule Set reads, at most `SWWAF_WAF_BODY_LIMIT` and one byte
+  more, which is held until the app is sent it. A WebSocket, or any other
   upgraded connection, passes through, and the timeouts do not cut it.
 - Works out the client's address. A TCP peer outside `SWWAF_TRUSTED_PROXIES` is
   the client, and the forwarded headers it sends are replaced, not passed on.
@@ -196,33 +197,41 @@ in `bin/state` unless `SWWAF_STATE_DIR` is set, and the default rule file of
   rate limits and before the rule files, which it does not need. A request for
   one is refused with `SWWAF_BAN_RESPONSE` and bans the client's netblock for a
   clear sign of attack, as a `ban` rule's match does. A trap path is matched
-  against the path a `path` rule sees: as the client sent it, before any
+  against the path a `path` rule sees: without the query, as the client sent it
-  decoding and without the query, the whole of it, character for character.
+  and once percent-decoded, the whole of it, character for character.
-  `/wp-login.php` matches `/wp-login.php?redirect_to=x`, but not
+  `/wp-login.php` matches `/wp-login.php?redirect_to=x` and `/%77p-login.php`,
-  `/wp-login.php/`, `/WP-LOGIN.PHP`, `/blog/wp-login.php` or `/%77p-login.php`.
+  but not `/wp-login.php/`, `/WP-LOGIN.PHP`, `/blog/wp-login.php` or
-  A client in `SWWAF_ALLOW_NETS` is not checked.
+  `/%2577p-login.php`. The ban's notes give the path as the client sent it. A
+  client in `SWWAF_ALLOW_NETS` is not checked.
 - Inspects each request with the OWASP Core Rule Set 4.25.0, run by Coraza,
   after the rule files, unless `SWWAF_WAF_MODE` is `off`: its method, its URL
-  with the query, and its headers, but no request body and no response. Each of
+  with the query, and its headers, and, once `SWWAF_WAF_BODY_LIMIT` is set, its
-  its rules that matches adds to the request's anomaly score, up to the paranoia
+  body when it is form data, multipart, JSON or XML, as that setting below
-  level `SWWAF_WAF_PARANOIA_LEVEL` sets. A request with more than 1000 query
+  describes; no response is inspected. Each of its rules that matches adds to
-  parameters adds 5 (rule 900300), as a rule rated critical does, since Coraza
+  the request's anomaly score, up to the paranoia level
-  reads only the first 1000. A score at or over `SWWAF_WAF_ANOMALY_THRESHOLD`, 5
+  `SWWAF_WAF_PARANOIA_LEVEL` sets. A request with more than 1000 query
-  by default, is a match: in `block` mode, the default, the request is refused
+  parameters, or more than 1000 fields in a form data or JSON body it reads,
-  with `403`, and in `detect` mode it goes on to the app. Either way its log
+  adds 5 (rule 900300), as a rule rated critical does, since Coraza reads only
-  line names the rules and the score (see `waf_rule_ids` and `waf_score` in
+  the first 1000. A score at or over `SWWAF_WAF_ANOMALY_THRESHOLD`, 5 by
-  "Request log" below), and it raises a `waf_block` alert. A refusal bans no one
+  default, is a match: in `block` mode, the default, the request is refused with
-  by itself, since the Core Rule Set takes some ordinary requests for attacks,
+  `403`, and in `detect` mode it goes on to the app. Either way its log line
-  but it is an offence the client's history counts, and it counts toward the
+  names the rules and the score (see `waf_rule_ids` and `waf_score` in "Request
-  error burst; a match in `detect` mode is neither. A request a rule file
+  log" below), and it raises a `waf_block` alert. A refusal bans no one by
-  refuses, one for a path `SWWAF_WAF_EXEMPT_PATHS` exempts, and one from a
+  itself, since the Core Rule Set takes some ordinary requests for attacks, but
-  client in `SWWAF_ALLOW_NETS` are not inspected. `smallwebwaf` changes the Core
+  it is an offence the client's history counts, and it counts toward the error
-  Rule Set in six ways, so that gitea's ordinary requests get through, and no
+  burst; a match in `detect` mode is neither. A request a rule file refuses, one
-  setting undoes them:
+  for a path `SWWAF_WAF_EXEMPT_PATHS` exempts, and one from a client in
+  `SWWAF_ALLOW_NETS` are not inspected. `smallwebwaf` changes the Core Rule Set
+  in six ways, so that gitea's ordinary requests get through, and no setting
+  undoes them:
     - `PUT`, `PATCH` and `DELETE` are allowed besides `GET`, `HEAD`, `POST` and
       `OPTIONS`; any other method stays refused.
     - The headers `Expect` and `Content-Encoding` are allowed; the others the
       Core Rule Set refuses, such as `Proxy` and `Content-Range`, stay refused.
+      Once `SWWAF_WAF_BODY_LIMIT` is set, `Content-Encoding` is refused again
+      (rule 920450) on form data, multipart, JSON and XML, the kinds of body the
+      Core Rule Set reads, since a compressed body cannot be inspected; so it is
+      on a JSON or XML body too large to be read.
     - The query parameter `redirect_uri` is not checked for a URL naming an IP
       address or `localhost` (rules 931100 and 934110), which Git Credential
       Manager, git-credential-oauth and tea ask to be sent back to.
@@ -232,7 +241,14 @@ in `bin/state` unless `SWWAF_STATE_DIR` is set, and the default rule file of
       command names (932260), so that a file such as `.gitignore` or a branch
       such as `docker-build` gets through there. So does what only these rules
       refuse, such as `|cat /etc/passwd`; path traversal and SQL injection are
-      still refused there.
+      still refused there. These names, and `redirect_uri` above, are matched
+      without regard to case, as Coraza matches them, so `Path` or `PATH` is
+      treated as `path`. Once `SWWAF_WAF_BODY_LIMIT` is set, they are left out
+      in the same way among the fields of a form data or multipart body, which
+      Coraza holds with the query parameters, and gitea posts some of them in
+      its forms: `redirect_uri` when an OAuth sign-in is granted, and `ref` when
+      a workflow is run by hand. A field of a JSON body is named by its path,
+      such as `json.path`, and keeps these rules.
     - The cookies `gitea_flash` and `redirect_to` are not read, and `Referer` is
       not checked for a Unix command given without arguments (932340) or for
       Java starting a process (944110); it is checked by every other rule.
@@ -695,12 +711,39 @@ effective settings are logged at start, unless `SWWAF_LOG_LEVEL` is `warn` or
   Core Rule Set does not inspect, each starting with `/`, matched as
   `SWWAF_RATE_LIMIT_EXEMPT_PATHS` matches its own: a request whose path holds
   `..`, a backslash or an encoded slash is inspected whatever its prefix.
+- `SWWAF_WAF_BODY_LIMIT` (default `off`): `off` has the Core Rule Set read no
+  request body. A size, such as `128K`, at most `1G`, has it read a body of form
+  data or multipart up to that size, the rest of a longer one passing on to the
+  app as it arrives, without being held, and a JSON or XML body no larger than
+  that size, since those cannot be read in part. A body is JSON when its type is
+  `application/json` or `text/json`, or an `application/` or `text/` type ending
+  in `+json`, and XML when its type is `application/xml` or `text/xml`, or an
+  `application/` or `text/` type ending in `+xml`. Any other body reaches the
+  app uninspected, and so does a larger JSON or XML body: the Core Rule Set
+  would read any other body as form data, where binary content such as a git
+  push trips rules written for text. A body it reads that Coraza cannot parse
+  (rule 900440), and a multipart body that fails Coraza's strict checks (rule
+  900450), add 5 to the score, as a rule rated critical does, since no rule
+  reads what comes after the fault. So does a multipart body the limit cuts
+  before the colon of a part's header line, or between the carriage return and
+  the line feed that end a part's header line or the empty line after its
+  headers, since Coraza takes the line the limit cuts for a malformed header.
+  The client has until `SWWAF_CLIENT_REQUEST_TIMEOUT` runs out to send the part
+  that is read, and a request whose body passes `SWWAF_REQUEST_MAX_BYTES` within
+  it is refused before anything reaches the app. Of a file in a multipart body,
+  Coraza counts the bytes and writes nothing. Body inspection suits apps whose
+  forms carry no code. In front of gitea it refuses issue and comment text, wiki
+  pages and files saved in the web editor that hold shell commands or code
+  (932125, 932235, 932250 and others), package descriptions that show code, PyPI
+  uploads (922130), and attachments named like `debug.log` or `config.yml`
+  (932180), until the rule ids the request log names are added to
+  `SWWAF_WAF_DISABLED_RULES`.
 - `SWWAF_TRAP_PATHS` (default empty): paths the app never serves and only
   scanners ask for, such as `/wp-login.php,/xmlrpc.php` in front of gitea; a
   request for one bans its client for a clear sign of attack, as a `ban` rule
   does, with or without rule files (see "What it does so far" above). A path
-  that does not start with `/`, or holds a `?`, which no path a `path` rule sees
+  that does not start with `/`, or holds a `?`, as one written with a query
-  holds, stops the start.
+  does, stops the start.
 - `SWWAF_ERROR_BURST_THRESHOLD` (default `30`): the most requests of a client in
   a minute that `smallwebwaf` may refuse after a rule file match, a trap path or
   a Core Rule Set match, or for a missing or wrong token; one more breaks the
@@ -983,13 +1026,13 @@ which every line has.
   from when the request's headers had been read to when its line is written, and
   `duration_checks` over the same start to when the checks were done; the health
   check runs none, and its line has no `duration_checks`. `duration_waf`, the
-  part of the checks the Core Rule Set took, is there with `waf_score`.
+  part of the checks the Core Rule Set took, reading the part of the body it
-  `duration_upstream_connect`, `duration_upstream_first_byte` and
+  reads included, is there with `waf_score`. `duration_upstream_connect`,
-  `duration_upstream_total` are there for a request passed to the app, and run
+  `duration_upstream_first_byte` and `duration_upstream_total` are there for a
-  from when it was handed to the app: until there was a connection to it, new or
+  request passed to the app, and run from when it was handed to the app: until
-  kept open from an earlier request, until the first byte of its answer arrived,
+  there was a connection to it, new or kept open from an earlier request, until
-  and until the end. The first two are left out when that never happened, as for
+  the first byte of its answer arrived, and until the end. The first two are
-  an app that cannot be reached.
+  left out when that never happened, as for an app that cannot be reached.
 
 No body is logged, and no header but those above. `smallwebwaf`'s own messages
 (start, the settings, stop, errors) share the stream as JSON lines marked
@@ -1480,14 +1523,15 @@ scanner-agent  user_agent  ban     (?i)\b(sqlmap|nikto|nuclei|wpscan)\b
 
 - The id is letters, digits, `-` and `_`, and no two rules share one. The
   request log, the metrics and a ban's notes name the rule by it.
-- The target is what the regex is matched against: `path` or `query`, as the
+- The target is what the regex is matched against: `path`, the path without the
-  client sent it, before any decoding; `uri`, the path and the query together,
+  query, and `uri`, the path and the query together, each both as sent and once
-  both as sent and once percent-decoded, so that an encoded probe does not slip
+  percent-decoded, so that an encoded probe does not slip past; `query`, as the
-  past; `method`; `host`; `user_agent`; `referer`; or `header:<Name>`, any one
+  client sent it, before any decoding; `method`; `host`; `user_agent`;
-  request header but `Host` and `Transfer-Encoding`, which Go's HTTP server
+  `referer`; or `header:<Name>`, any one request header but `Host` and
-  takes out of every request; the request's host is the target `host`. A header
+  `Transfer-Encoding`, which Go's HTTP server takes out of every request; the
-  sent more than once is matched with its values joined by `, `, and one not
+  request's host is the target `host`. A header sent more than once is matched
-  sent as empty text. No body is read.
+  with its values joined by `, `, and one not sent as empty text. No body is
+  read.
 - The action is `log`, `block` or `ban` (see "What it does so far" above). Keep
   `ban` for requests no real visitor sends, and anchor a path at the site root
   with `^/`: a file of the same name deeper in a site can be ordinary content,
@@ -2212,15 +2256,17 @@ alerts, the metrics nor `reputation.json` hold it. Given as a file, with
   which bans the client, for a DNSBL zone's verdict, for AbuseIPDB's score, for
   a rate limit, which bans the client, for a trap path, which bans the client,
   for a `block` or `ban` rule, the latter banning the client, for a Core Rule
-  Set match in `block` mode, and for an announced body over the size limit; in
+  Set match in `block` mode, for a body that passes the size limit or
-  `observe` mode, only for the size limit, with what it would have refused for
+  `SWWAF_CLIENT_REQUEST_TIMEOUT` while the Core Rule Set reads it, and for an
-  noted in the log line. A request under `/_smallwebwaf/` that `check` lets
+  announced body over the size limit; in `observe` mode, only for the last two,
-  through is answered by `answerAdmin` instead of reaching the app. Once the
+  with what it would have refused for noted in the log line. A request under
-  answer to a request passed to the app has ended, `countBytes` counts its bytes
+  `/_smallwebwaf/` that `check` lets through is answered by `answerAdmin`
-  for the byte limits, and once any request but the health check has ended,
+  instead of reaching the app. Once the answer to a request passed to the app
-  `countRefusal` counts it for the error burst if it was refused after a rule
+  has ended, `countBytes` counts its bytes for the byte limits, and once any
-  file match, a trap path or a Core Rule Set match or for its token, and
+  request but the health check has ended, `countRefusal` counts it for the error
-  `countAnomalies` counts it for the anomaly thresholds.
+  burst if it was refused after a rule file match, a trap path or a Core Rule
+  Set match or for its token, and `countAnomalies` counts it for the anomaly
+  thresholds.
 - `internal/metrics`: the metrics, counted as the other parts tell it what
   happened, and served in the Prometheus text format.
 - `internal/bans`: the ban ledger: each netblock's bans with their notes, how
@@ -2229,8 +2275,8 @@ alerts, the metrics nor `reputation.json` hold it. Given as a file, with
 - `internal/rules`: reads the rule files at start and again as they change, and
   tells which of their rules a request matches.
 - `internal/waf`: the Core Rule Set with the six changes, as Coraza's own
-  directives, and what it finds in a request's method, URL and headers: the
+  directives, and what it finds in a request's method, URL, headers and the part
-  rules that matched and the anomaly score.
+  of its body it reads: the rules that matched and the anomaly score.
 - `internal/lookup`: looks up each client's AS number and country through GeoJS,
   keeps the answers, and hands each new one to the proxy, which adds it to the
   client's history and to the notes of its bans; or in the lookup database,

SPEC.md

@@ -618,10 +618,12 @@ The settings, by group:
           `|cat /etc/passwd`, `wget http://…` and `nc -e /bin/sh …`, and
           `file:///etc/passwd`, pass as well. Path traversal (`../`), SQL and
           script injection and PHP, Java and Node.js code are still refused
-          there, and every other parameter keeps all three rules. An app that
+          there, and every other parameter keeps all three rules. These names,
-          uses one of these parameters as a file on the server, or passes it to
+          and `redirect_uri` in the change before, are matched without regard to
-          a shell, gets no help from the three rules there (see "Risks the
+          case, as Coraza matches them, so `Path` or `PATH` is treated as
-          design has to handle").
+          `path`. An app that uses one of these parameters as a file on the
+          server, or passes it to a shell, gets no help from the three rules
+          there (see "Risks the design has to handle").
         - The Core Rule Set reads the request without the `gitea_flash` and
           `redirect_to` cookies, and does not check `Referer` for a Unix command
           given without arguments (932340) or for Java starting a process
@@ -823,7 +825,8 @@ a file edited, added or removed there takes effect while it runs.
 - `id`: a short name of letters, digits, `-` and `_`, unique across all files.
   It appears in the request log, metrics, alerts and ban notes.
 - `target`: what the regex is matched against.
-    - `path`: the URL path as received, before any decoding.
+    - `path`: the URL path without the query, both as received and once
+      percent-decoded, so an encoded probe cannot slip past.
     - `query`: the raw query string.
     - `uri`: path and query together, both as received and once percent-decoded,
       so an encoded probe cannot slip past.

internal/bans/bans.go

@@ -123,7 +123,8 @@ type Notes struct {
 	LimitPercentSetting string `json:"limit_percent_setting,omitempty"`
 	// RuleID and Target are, for a ban for a clear sign of attack, the id
 	// of the rule file rule that matched, and its target; TrapPath is, for
-	// one for a request for a path in SWWAF_TRAP_PATHS, that path.
+	// one for a request for a path in SWWAF_TRAP_PATHS, that path as the
+	// client sent it.
 	RuleID   string `json:"rule_id,omitempty"`
 	Target   string `json:"target,omitempty"`
 	TrapPath string `json:"trap_path,omitempty"`

internal/config/config.go

@@ -244,14 +244,16 @@ type Config struct {
 	// level, from 1 to 4 (SWWAF_WAF_PARANOIA_LEVEL), and
 	// WAFAnomalyThreshold the anomaly score at which a request is a match
 	// (SWWAF_WAF_ANOMALY_THRESHOLD), 0 while it is off. WAFDisabledRules
-	// are the ids of its rules switched off (SWWAF_WAF_DISABLED_RULES), and
+	// are the ids of its rules switched off (SWWAF_WAF_DISABLED_RULES),
 	// WAFExemptPaths the path prefixes it does not inspect
-	// (SWWAF_WAF_EXEMPT_PATHS).
+	// (SWWAF_WAF_EXEMPT_PATHS), and WAFBodyLimit the most of a request body
+	// it reads (SWWAF_WAF_BODY_LIMIT), 0 while it is off and it reads none.
 	WAFMode             string
 	WAFParanoiaLevel    int
 	WAFAnomalyThreshold int
 	WAFDisabledRules    []int
 	WAFExemptPaths      []string
+	WAFBodyLimit        int64
 	// TrapPaths are the paths a request for which is a clear sign of
 	// attack (SWWAF_TRAP_PATHS), each starting with / and without a ?.
 	TrapPaths []string
@@ -396,6 +398,7 @@ var (
 	errNeedsDBPath          = errors.New("it names the file to look clients up in")
 	errDBPathUnused         = errors.New("only file reads it")
 	errNotOver4K            = errors.New("is not a size of more than 4K, such as 32K")
+	errOver1G               = errors.New("is more than 1G, the most Coraza reads")
 	errNotDurationAboveZero = errors.New(
 		"is not a duration above zero, such as 1h or 7d")
 	errNotNumberAboveZero = errors.New(
@@ -559,6 +562,7 @@ func FromEnvironment(lookupEnv func(string) (string, bool)) (*Config, error) {
 		WAFDisabledRules: env.ruleIDs("SWWAF_WAF_DISABLED_RULES",
 			"920340,920420,920440,920640,930130,930140"),
 		WAFExemptPaths:       env.pathPrefixes("SWWAF_WAF_EXEMPT_PATHS", ""),
+		WAFBodyLimit:         env.wafBodyLimit("SWWAF_WAF_BODY_LIMIT", off),
 		TrapPaths:            env.trapPaths("SWWAF_TRAP_PATHS"),
 		ErrorBurstThreshold:  env.count("SWWAF_ERROR_BURST_THRESHOLD", "30"),
 		LogRemoteURL:         env.logRemoteURL("SWWAF_LOG_REMOTE_URL"),
@@ -764,6 +768,15 @@ func (e *environment) size(name, defaultValue string) int64 {
 	return size
 }
 
+// wafBodyLimit reads the setting that is the most of a request body the
+// Core Rule Set reads.
+func (e *environment) wafBodyLimit(name, defaultValue string) int64 {
+	limit, err := parseWAFBodyLimit(e.value(name, defaultValue))
+	e.check(name, err)
+
+	return limit
+}
+
 // headerSize reads the setting that is the largest request line and
 // headers.
 func (e *environment) headerSize(name, defaultValue string) int64 {
@@ -1432,6 +1445,22 @@ func parseHeaderSize(value string) (int64, error) {
 	return size, nil
 }
 
+// parseWAFBodyLimit reads the most of a request body the Core Rule Set
+// reads: a size as parseSize reads it, or off, but at most 1G, since
+// Coraza, which runs the Core Rule Set, refuses to load with more.
+func parseWAFBodyLimit(value string) (int64, error) {
+	limit, err := parseSize(value)
+	if err != nil {
+		return 0, err
+	}
+
+	if limit > gibibyte {
+		return 0, fmt.Errorf("%q %w", value, errOver1G)
+	}
+
+	return limit, nil
+}
+
 // splitUnit splits a size into its number and the bytes its suffix
 // stands for.
 func splitUnit(value string) (string, int64) {
@@ -1636,7 +1665,8 @@ func parsePathPrefixes(value string) ([]string, error) {
 
 // parseTrapPaths reads a comma-separated list of trap paths. Each is
 // matched against a request's path as a path rule is, without the query,
-// so a path that does not start with / or holds a ? would never match.
+// so a path that does not start with / would never match, and one that
+// holds a ? is written with a query.
 func parseTrapPaths(value string) ([]string, error) {
 	paths, err := parseList(value)
 	if err != nil {

internal/config/config_test.go

@@ -96,6 +96,7 @@ const (
 	wafAnomalyThreshold     = "SWWAF_WAF_ANOMALY_THRESHOLD"
 	wafDisabledRules        = "SWWAF_WAF_DISABLED_RULES"
 	wafExemptPaths          = "SWWAF_WAF_EXEMPT_PATHS"
+	wafBodyLimit            = "SWWAF_WAF_BODY_LIMIT"
 	trapPaths               = "SWWAF_TRAP_PATHS"
 	errorBurstThreshold     = "SWWAF_ERROR_BURST_THRESHOLD"
 	logRemoteURL            = "SWWAF_LOG_REMOTE_URL"
@@ -580,15 +581,20 @@ func TestCoreRuleSetSettings(t *testing.T) {
 			environment{
 				wafMode: config.WAFModeDetect, wafParanoiaLevel: "4", wafAnomalyThreshold: "10",
 				wafDisabledRules: "942100, 920350", wafExemptPaths: "/api/, /static/",
+				wafBodyLimit: "128K",
 			},
 			config.Config{
 				WAFMode: config.WAFModeDetect, WAFParanoiaLevel: 4, WAFAnomalyThreshold: 10,
 				WAFDisabledRules: []int{942100, 920350},
 				WAFExemptPaths:   []string{"/api/", "/static/"},
+				WAFBodyLimit:     128 << 10,
 			},
 		},
 		{
-			environment{wafMode: off, wafAnomalyThreshold: off, wafDisabledRules: ""},
+			environment{
+				wafMode: off, wafAnomalyThreshold: off, wafDisabledRules: "",
+				wafBodyLimit: off,
+			},
 			config.Config{
 				WAFMode: config.WAFModeOff, WAFParanoiaLevel: 1, WAFAnomalyThreshold: 0,
 				WAFDisabledRules: []int{}, WAFExemptPaths: []string{},
@@ -601,6 +607,7 @@ func TestCoreRuleSetSettings(t *testing.T) {
 			WAFMode: cfg.WAFMode, WAFParanoiaLevel: cfg.WAFParanoiaLevel,
 			WAFAnomalyThreshold: cfg.WAFAnomalyThreshold,
 			WAFDisabledRules:    cfg.WAFDisabledRules, WAFExemptPaths: cfg.WAFExemptPaths,
+			WAFBodyLimit: cfg.WAFBodyLimit,
 		}
 		if !reflect.DeepEqual(got, tc.want) {
 			t.Errorf("%v gave\n%+v\nwant\n%+v", tc.env, got, tc.want)
@@ -608,6 +615,15 @@ func TestCoreRuleSetSettings(t *testing.T) {
 	}
 }
 
+func TestWAFBodyLimitOf1G(t *testing.T) {
+	t.Parallel()
+
+	cfg := fromEnvironment(t, environment{wafBodyLimit: "1G"})
+	if cfg.WAFBodyLimit != 1<<30 {
+		t.Errorf("1G read as %d", cfg.WAFBodyLimit)
+	}
+}
+
 func TestInvalidCoreRuleSetSettingStopsTheStart(t *testing.T) {
 	t.Parallel()
 
@@ -636,16 +652,28 @@ func TestInvalidCoreRuleSetSettingStopsTheStart(t *testing.T) {
 			`"-942100" is not the id of a Core Rule Set rule, ` +
 				`a whole number such as 942100`,
 		},
-		// The paranoia level, the allowed methods, the headers refused, and
+		// The paranoia level, the allowed methods, the headers refused, a
-		// a request with more query parameters than Coraza keeps.
+		// request with more query parameters than Coraza keeps, and a body
+		// Coraza cannot parse or that fails its strict checks.
 		{wafDisabledRules, "942100,900000", `"900000"` + setupRule},
 		{wafDisabledRules, "942100,900200", `"900200"` + setupRule},
 		{wafDisabledRules, "942100,900250", `"900250"` + setupRule},
 		{wafDisabledRules, "942100,900300", `"900300"` + setupRule},
+		{wafDisabledRules, "942100,900440", `"900440"` + setupRule},
+		{wafDisabledRules, "942100,900450", `"900450"` + setupRule},
 		{
 			wafExemptPaths, "api/",
 			`"api/" is not a path prefix starting with /, such as /assets/`,
 		},
+		{
+			wafBodyLimit, "128KB",
+			`"128KB" is not a size such as 512K, 100M or 5G, or off`,
+		},
+		{wafBodyLimit, "2G", `"2G" is more than 1G, the most Coraza reads`},
+		{
+			wafBodyLimit, "1073741825",
+			`"1073741825" is more than 1G, the most Coraza reads`,
+		},
 	} {
 		t.Run(tc.name+"="+tc.value, func(t *testing.T) {
 			t.Parallel()
@@ -2403,6 +2431,7 @@ func TestLogsEachSettingWithItsValue(t *testing.T) {
 		wafAnomalyThreshold:     "5",
 		wafDisabledRules:        defaultWAFDisabledRules,
 		wafExemptPaths:          "",
+		wafBodyLimit:            off,
 		trapPaths:               "",
 		errorBurstThreshold:     "30",
 		logRemoteURL:            "",

internal/proxy/bodies.go

@@ -21,6 +21,9 @@ type requestBody struct {
 	// SWWAF_REQUEST_MAX_BYTES.
 	body io.ReadCloser
 	rq   *request
+	// readByCoreRuleSet is what the Core Rule Set read of the body before
+	// the request went to the app, and Read gives first.
+	readByCoreRuleSet []byte
 	// waiting is true while a Read waits for the client to send more.
 	waiting atomic.Bool
 	// received is true once the client has sent the whole body.
@@ -29,8 +32,16 @@ type requestBody struct {
 	bytes atomic.Int64
 }
 
-// Read reads from the client's body.
+// Read reads from the client's body, after what the Core Rule Set read of
+// it, which has been counted already.
 func (b *requestBody) Read(p []byte) (int, error) {
+	if len(b.readByCoreRuleSet) > 0 {
+		n := copy(p, b.readByCoreRuleSet)
+		b.readByCoreRuleSet = b.readByCoreRuleSet[n:]
+
+		return n, nil
+	}
+
 	b.waiting.Store(true)
 	n, err := b.body.Read(p)
 	b.waiting.Store(false)

internal/proxy/coreruleset.go

@@ -1,6 +1,9 @@
 package proxy
 
 import (
+	"errors"
+	"net/http"
+	"os"
 	"time"
 
 	"sneak.berlin/go/smallwebwaf/internal/alerts"
@@ -16,7 +19,8 @@ import (
 // SWWAF_WAF_ANOMALY_THRESHOLD is a match: it raises the waf_block alert,
 // and in block mode refuses the request, which is an offence its client's
 // history counts, and so returns ActionWAFBlocked. It returns "" for a
-// request it does not refuse.
+// request it does not refuse, and for one whose body meets a size or time
+// limit while the Core Rule Set reads it, which it notes nothing of.
 func (rq *request) checkCoreRuleSet() string {
 	cfg := rq.h.config
 	if cfg.WAFMode == config.WAFModeOff || pathExempt(rq.in.URL, cfg.WAFExemptPaths) {
@@ -24,7 +28,12 @@ func (rq *request) checkCoreRuleSet() string {
 	}
 
 	start := time.Now()
-	result := rq.h.coreRuleSet.Inspect(rq.in, rq.client)
+
+	result := rq.inspect()
+	if rq.refused.Load() != nil {
+		return "" // the refusal for that limit, which check returns
+	}
+
 	rq.line.DurationWAF = new(requestlog.Milliseconds(time.Since(start)))
 	rq.line.WAFRuleIDs = result.RuleIDs
 	rq.line.WAFScore = &result.Score
@@ -49,6 +58,39 @@ func (rq *request) checkCoreRuleSet() string {
 	return requestlog.ActionWAFBlocked
 }
 
+// inspect runs the Core Rule Set on the request, which reads the part of
+// its body it inspects within SWWAF_CLIENT_REQUEST_TIMEOUT, and keeps that
+// part for the app. A client that runs out of time is refused with 408
+// here, and a body over SWWAF_REQUEST_MAX_BYTES with 413 as it is read;
+// check returns the refusal. A body that breaks off for any other reason
+// is passed on as far as it came, and the request to the app fails there,
+// as it would have without the Core Rule Set.
+func (rq *request) inspect() waf.Result {
+	if rq.body == nil {
+		// Nothing is read of no body, so nothing can go wrong reading it.
+		result, _, _ := rq.h.coreRuleSet.Inspect(rq.in, rq.client, http.NoBody)
+
+		return result
+	}
+
+	_ = rq.rc.SetReadDeadline(rq.clientRequestDeadline())
+	result, read, err := rq.h.coreRuleSet.Inspect(rq.in, rq.client, rq.body)
+	// The timeouts that run while the request goes to the app take over.
+	_ = rq.rc.SetReadDeadline(time.Time{})
+
+	rq.body.readByCoreRuleSet = read
+
+	if errors.Is(err, os.ErrDeadlineExceeded) {
+		rq.refuse(refusal{
+			status: http.StatusRequestTimeout,
+			action: requestlog.ActionTimedOut,
+			limit:  "SWWAF_CLIENT_REQUEST_TIMEOUT",
+		})
+	}
+
+	return result
+}
+
 // alertWAFBlock raises the waf_block alert for the request, which the Core
 // Rule Set scored at result, at or over SWWAF_WAF_ANOMALY_THRESHOLD. Its
 // detail gives the rule ids, the score, the method and the path with the

internal/proxy/coreruleset_test.go

@@ -1,11 +1,14 @@
 package proxy_test
 
 import (
+	"io"
 	"net/http"
 	"net/netip"
 	"slices"
+	"strconv"
 	"strings"
 	"testing"
+	"time"
 
 	"sneak.berlin/go/smallwebwaf/internal/alerts"
 	"sneak.berlin/go/smallwebwaf/internal/ratelimit"
@@ -18,10 +21,14 @@ const (
 	wafAnomalyThreshold = "SWWAF_WAF_ANOMALY_THRESHOLD"
 	wafDisabledRules    = "SWWAF_WAF_DISABLED_RULES"
 	wafExemptPaths      = "SWWAF_WAF_EXEMPT_PATHS"
+	wafBodyLimit        = "SWWAF_WAF_BODY_LIMIT"
 	block               = "block"
 	detect              = "detect"
 )
 
+// formData is the type of a form's body.
+const formData = "application/x-www-form-urlencoded"
+
 // sqlInjection asks for / with an SQL injection in its query, which only
 // the Core Rule Set's rule 942100 matches, with a score of 5, the default
 // SWWAF_WAF_ANOMALY_THRESHOLD.
@@ -224,6 +231,197 @@ func TestDisabledRulesSwitchOffWhatGiteaWouldBeRefused(t *testing.T) {
 	}
 }
 
+func TestAttackInAFormBodyIsRefusedOnlyWhileBodiesAreRead(t *testing.T) {
+	t.Parallel()
+
+	const body = "id=1'%20OR%20'1'='1"
+
+	header := "Content-Type: " + formData + "\r\nContent-Length: " +
+		strconv.Itoa(len(body))
+
+	s, _, _ := startWithClock(t, "", map[string]string{wafMode: block})
+	line, _ := s.requestWithBody(http.MethodPost, client, "/", header, body,
+		http.StatusOK, requestlog.ActionForward)
+	wantWAF(t, line, new(0))
+
+	s, _, _ = startWithClock(t, "", map[string]string{
+		wafMode: block, wafBodyLimit: sizeLimitSetting,
+	})
+	line, _ = s.requestWithBody(http.MethodPost, client, "/", header, body,
+		http.StatusForbidden, requestlog.ActionWAFBlocked)
+	wantWAF(t, line, new(5), 942100)
+}
+
+func TestBodiesReachTheAppAsSentWhileBodiesAreRead(t *testing.T) {
+	t.Parallel()
+
+	// The app answers with the body it was sent, once it has the whole of
+	// it: Go's server reads no more of a body once the answer has begun.
+	app := startApp(t, func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		_, _ = w.Write(body)
+	})
+	addr, out := startProxy(t, app.URL, map[string]string{
+		wafMode: block, wafBodyLimit: sizeLimitSetting,
+	})
+	longer := "a=" + strings.Repeat("b", 64*sizeLimit)
+
+	for i, tc := range []struct {
+		name, contentType, body string
+		// announced sends the body's length in Content-Length; otherwise
+		// the body is sent in chunks with no length given.
+		announced bool
+	}{
+		{"form data within the limit", formData, "a=b", true},
+		{"form data longer than the limit", formData, longer, true},
+		{"form data longer than the limit, not announced", formData, longer, false},
+		{
+			"JSON larger than the limit", "application/json",
+			`{"a":"` + strings.Repeat("b", 2*sizeLimit) + `"}`, true,
+		},
+		{
+			"a binary body", "application/octet-stream",
+			strings.Repeat("\x00\xff", sizeLimit), true,
+		},
+	} {
+		// A reader whose length the client cannot tell is sent in chunks.
+		var body io.Reader = strings.NewReader(tc.body)
+		if !tc.announced {
+			body = io.MultiReader(body)
+		}
+
+		req := newRequest(t, http.MethodPost, addr, "/", body)
+		req.Header.Set("Content-Type", tc.contentType)
+
+		got := do(t, req)
+		if got.status != http.StatusOK || string(got.body) != tc.body {
+			t.Errorf("%s: the app got %d bytes, answered %d, want the %d sent, 200",
+				tc.name, len(got.body), got.status, len(tc.body))
+		}
+
+		line := out.requestLines(t, i+1)[i]
+		wantLine(t, line, http.StatusOK, requestlog.ActionForward)
+
+		if line.RequestBytes != int64(len(tc.body)) {
+			t.Errorf("%s: log line has request_bytes %d, want %d", tc.name,
+				line.RequestBytes, len(tc.body))
+		}
+	}
+}
+
+func TestFormBodyLongerThanTheLimitStreamsOnToTheApp(t *testing.T) {
+	t.Parallel()
+
+	const (
+		first = "a=" // and twice the limit of b's, then the rest
+		rest  = 64 * sizeLimit
+	)
+
+	// past is closed once the app has received twice what the Core Rule
+	// Set reads, and got is the length of the whole body it received.
+	past := make(chan struct{})
+	got := make(chan int64, 1)
+	app := startApp(t, func(_ http.ResponseWriter, r *http.Request) {
+		n, _ := io.CopyN(io.Discard, r.Body, 2*sizeLimit)
+
+		close(past)
+
+		m, _ := io.Copy(io.Discard, r.Body)
+		got <- n + m
+	})
+	addr, out := startProxy(t, app.URL, map[string]string{
+		wafMode: block, wafBodyLimit: sizeLimitSetting,
+	})
+
+	// The client sends the rest only once the app has received the first
+	// part: were smallwebwaf to hold the body until the end, it would
+	// never come.
+	body, sender := io.Pipe()
+
+	go func() {
+		_, _ = io.WriteString(sender, first+strings.Repeat("b", 2*sizeLimit))
+
+		select {
+		case <-past:
+		case <-time.After(waitLimit):
+			t.Error("the app got no more than the Core Rule Set reads " +
+				"before the whole body was sent")
+
+			_ = sender.CloseWithError(io.ErrUnexpectedEOF)
+
+			return
+		}
+
+		_, _ = io.WriteString(sender, strings.Repeat("b", rest))
+		_ = sender.Close()
+	}()
+
+	req := newRequest(t, http.MethodPost, addr, "/", body)
+	req.Header.Set("Content-Type", formData)
+	wantStatus(t, do(t, req), http.StatusOK)
+
+	want := int64(len(first) + 2*sizeLimit + rest)
+	if n := <-got; n != want {
+		t.Errorf("the app got %d bytes, want %d", n, want)
+	}
+
+	wantLine(t, out.requestLine(t), http.StatusOK, requestlog.ActionForward)
+}
+
+func TestClientTooSlowToSendWhatTheCoreRuleSetReads(t *testing.T) {
+	t.Parallel()
+
+	app := startApp(t, func(http.ResponseWriter, *http.Request) {})
+	addr, out := startProxy(t, app.URL, map[string]string{
+		wafMode: block, wafBodyLimit: sizeLimitSetting,
+		clientRequestTimeout: shortTimeoutSetting, metricsToken: token,
+	})
+
+	conn := dial(t, addr)
+	send(t, conn, "POST /comment HTTP/1.1\r\nHost: app\r\nContent-Type: "+formData+
+		"\r\nContent-Length: 100\r\n\r\ncontent=the first bytes")
+
+	wantStatus(t, readResponse(t, conn), http.StatusRequestTimeout)
+
+	line := out.requestLine(t)
+	wantLine(t, line, http.StatusRequestTimeout, requestlog.ActionTimedOut)
+	wantNotSentToTheApp(t, line)
+	wantLimitHits(t, addr, clientRequestTimeout, 1)
+}
+
+func TestBodyOverTheSizeLimitWhileTheCoreRuleSetReadsIt(t *testing.T) {
+	t.Parallel()
+
+	app := startApp(t, func(http.ResponseWriter, *http.Request) {})
+	addr, out := startProxy(t, app.URL, map[string]string{
+		wafMode: block, wafBodyLimit: "4K",
+		requestMaxBytes: sizeLimitSetting, metricsToken: token,
+	})
+
+	// Sent in chunks, its length is not announced, and is found to be over
+	// the limit as the Core Rule Set reads it.
+	body := io.MultiReader(strings.NewReader("a=" + strings.Repeat("b", 2*sizeLimit)))
+	req := newRequest(t, http.MethodPost, addr, "/", body)
+	req.Header.Set("Content-Type", formData)
+	wantStatus(t, do(t, req), http.StatusRequestEntityTooLarge)
+
+	line := out.requestLine(t)
+	wantLine(t, line, http.StatusRequestEntityTooLarge, requestlog.ActionTooLarge)
+	wantNotSentToTheApp(t, line)
+	wantLimitHits(t, addr, requestMaxBytes, 1)
+}
+
+// wantNotSentToTheApp checks that the request of line was not sent to the
+// app at all.
+func wantNotSentToTheApp(t *testing.T, line logLine) {
+	t.Helper()
+
+	_, sent := line.fields["duration_upstream_total"]
+	if sent {
+		t.Error("log line has duration_upstream_total, for a request sent to the app")
+	}
+}
+
 func TestResponsesAreNotInspected(t *testing.T) {
 	t.Parallel()
 

internal/proxy/proxy.go

@@ -232,8 +232,8 @@ func newReputation(
 }
 
 // newCoreRuleSet returns the Core Rule Set at SWWAF_WAF_PARANOIA_LEVEL,
-// without the rules SWWAF_WAF_DISABLED_RULES switches off, or nil while
+// without the rules SWWAF_WAF_DISABLED_RULES switches off, reading bodies
-// SWWAF_WAF_MODE is off.
+// up to SWWAF_WAF_BODY_LIMIT, or nil while SWWAF_WAF_MODE is off.
 func newCoreRuleSet(cfg *config.Config) *waf.CoreRuleSet {
 	if cfg.WAFMode == config.WAFModeOff {
 		return nil
@@ -241,11 +241,12 @@ func newCoreRuleSet(cfg *config.Config) *waf.CoreRuleSet {
 
 	coreRuleSet, err := waf.New(waf.Params{
 		ParanoiaLevel: cfg.WAFParanoiaLevel, DisabledRules: cfg.WAFDisabledRules,
+		BodyLimit: cfg.WAFBodyLimit,
 	})
 	if err != nil {
 		// The Core Rule Set is built in, and the settings cannot break it:
-		// the paranoia level is from 1 to 4, and the id of no rule switches
+		// the paranoia level is from 1 to 4, the body limit at most 1G, and
-		// nothing off.
+		// the id of no rule switches nothing off.
 		panic(err)
 	}
 

internal/proxy/request.go

@@ -188,16 +188,23 @@ func requestHeaders(r *http.Request, names []string) map[string]string {
 }
 
 // check is the one place where a request can be refused once its client
-// is known, before its body is read or anything reaches the app. It
+// is known, before anything reaches the app, and before its body is read,
-// returns nil to let the request through. The checks of checkClient come
+// but for the part the Core Rule Set reads. It returns nil to let the
-// first, answered with SWWAF_BAN_RESPONSE, or 403 for a block rule or the
+// request through. The checks of checkClient come first, answered with
-// Core Rule Set, and then the size limit, so that a request the rate
+// SWWAF_BAN_RESPONSE, or 403 for a block rule or the Core Rule Set, and
-// limits count is counted even when it is refused for its size. In
+// then the size limit, so that a request the rate limits count is counted
-// observe mode a request checkClient refuses goes on to the size limit
+// even when it is refused for its size. In observe mode a request
-// like any other. ctx is the request's own context.
+// checkClient refuses goes on to the size limit like any other. A size or
+// time limit the Core Rule Set's reading of the body meets ends the
+// request in either mode. ctx is the request's own context.
 func (rq *request) check(ctx context.Context) *refusal {
 	action := rq.checkClient(ctx)
 
+	refused := rq.refused.Load()
+	if refused != nil {
+		return refused
+	}
+
 	switch {
 	case action == "":
 	case rq.h.config.Observe:
@@ -782,9 +789,19 @@ func (rq *request) responseTimedOut() {
 	}
 }
 
-// responseReceived is called once the app has sent its whole answer.
+// responseReceived is called once the app has sent its whole answer. An
+// app may answer before the client has sent its whole body, as with a 404
+// it sends without reading it. The timeouts then run on: until the
+// transport's read of the body returns, Go's server cannot send the
+// answer, and SWWAF_CLIENT_REQUEST_TIMEOUT is what ends that read for a
+// client that stops sending.
 func (rq *request) responseReceived() {
 	rq.complete = true
+
+	if rq.body != nil && !rq.body.received.Load() {
+		return
+	}
+
 	rq.stopTimers()
 }
 

internal/proxy/rules.go

@@ -10,13 +10,15 @@ import (
 )
 
 // trapPath reports whether the request asks for a path in
-// SWWAF_TRAP_PATHS: its path as a path rule sees it, before any decoding
+// SWWAF_TRAP_PATHS: its path as a path rule sees it, without the query,
-// and without the query, is one of them. Such a request is a clear sign of
+// as sent or once percent-decoded, is one of them. Such a request is a
-// attack, as a ban rule's match is: it bans the client's netblock, or in
+// clear sign of attack, as a ban rule's match is: it bans the client's
-// observe mode raises the alert for the ban it would have made.
+// netblock, or in observe mode raises the alert for the ban it would have
+// made. The ban's notes give the path as sent.
 func (rq *request) trapPath(now time.Time) bool {
 	path := rules.Path(rq.in)
-	if !slices.Contains(rq.h.config.TrapPaths, path) {
+	if !slices.Contains(rq.h.config.TrapPaths, path) &&
+		!slices.Contains(rq.h.config.TrapPaths, rules.DecodeOnce(path)) {
 		return false
 	}
 

internal/proxy/timeouts_test.go

@@ -194,6 +194,28 @@ func sendLargeBody(t *testing.T, addr string) net.Conn {
 	return conn
 }
 
+func TestClientStopsSendingAfterTheAppHasAnswered(t *testing.T) {
+	t.Parallel()
+
+	// The app answers at once without reading the body, as an app does
+	// for a path it does not serve.
+	app := startApp(t, http.NotFound)
+	addr, out := startProxy(t, app.URL, map[string]string{
+		clientRequestTimeout: shortTimeoutSetting,
+		metricsToken:         token,
+	})
+
+	start := time.Now()
+	got := readResponse(t, sendPartOfBody(t, addr))
+	wantTimedOut(t, start)
+
+	// The app's answer came before the time ran out, so the client is
+	// sent that answer, not smallwebwaf's 408.
+	wantStatus(t, got, http.StatusNotFound)
+	wantLine(t, out.requestLine(t), http.StatusNotFound, requestlog.ActionTimedOut)
+	wantLimitHits(t, addr, clientRequestTimeout, 1)
+}
+
 func TestAppTooSlowToAnswer(t *testing.T) {
 	t.Parallel()
 

internal/proxy/trappaths_test.go

@@ -32,10 +32,9 @@ func TestTrapPathBansAsABanRuleDoes(t *testing.T) {
 	})
 	start := clk.Now()
 
-	// Only the path itself, as the client sent it, is a trap path.
+	// Only the path itself, the whole of it in the same case, is a trap
-	for _, path := range []string{
+	// path.
-		"/wp-login.php/", "/WP-LOGIN.PHP", "/blog/xmlrpc.php", "/%77p-login.php",
+	for _, path := range []string{"/wp-login.php/", "/WP-LOGIN.PHP", "/blog/xmlrpc.php"} {
-	} {
 		s.request(otherClient, path, http.StatusOK, requestlog.ActionForward)
 	}
 
@@ -84,6 +83,22 @@ func TestTrapPathBansAsABanRuleDoes(t *testing.T) {
 	}
 }
 
+func TestTrapPathMatchesThePathDecodedOnce(t *testing.T) {
+	t.Parallel()
+
+	s, _, server := startWithClock(t, "", map[string]string{trapPaths: trapPathList})
+
+	// %2577 decodes once to %77, not to w.
+	s.request(otherClient, "/%2577p-login.php", http.StatusOK, requestlog.ActionForward)
+	s.request(client, "/%77p-login.php", http.StatusForbidden, requestlog.ActionBanned)
+
+	// The ban's notes give the path as the client sent it.
+	got := server.Ledger.Bans(netip.MustParsePrefix(client + "/32"))
+	if len(got) != 1 || got[0].Notes.TrapPath != "/%77p-login.php" {
+		t.Fatalf("bans %+v, want one whose notes give the trap path /%%77p-login.php", got)
+	}
+}
+
 func TestTrapPathsNeedNoRuleFiles(t *testing.T) {
 	t.Parallel()
 

internal/rules/rules.go

@@ -50,6 +50,13 @@ const quietTime = 2 * time.Second
 // header:<Name>.
 const headerTarget = "header:"
 
+// pathTarget and uriTarget are the targets matched both as received and
+// once percent-decoded: the path, and the path and query together.
+const (
+	pathTarget = "path"
+	uriTarget  = "uri"
+)
+
 // escapeLength is the length of a percent escape, such as %2e.
 const escapeLength = 3
 
@@ -374,7 +381,7 @@ func parse(line string) (Rule, bool, error) {
 // isTarget reports whether target is one a rule may have.
 func isTarget(target string) bool {
 	switch target {
-	case "path", "query", "uri", "method", "host", "user_agent", "referer":
+	case pathTarget, "query", uriTarget, "method", "host", "user_agent", "referer":
 		return true
 	}
 
@@ -384,35 +391,38 @@ func isTarget(target string) bool {
 }
 
 // matches reports whether the rule's regex matches its target in r. For
-// uri it is matched against the path and query as received, and against
+// path and uri it is matched against the target as received, and against
-// them once percent-decoded, so that an encoded probe cannot slip past.
+// it once percent-decoded, so that an encoded probe cannot slip past.
 func (rule Rule) matches(r *http.Request) bool {
-	if rule.Target == "uri" {
+	target := value(rule.Target, r)
-		uri := pathAndQuery(r)
 
-		return rule.regex.MatchString(uri) || rule.regex.MatchString(decodeOnce(uri))
+	if rule.Target == pathTarget || rule.Target == uriTarget {
+		return rule.regex.MatchString(target) || rule.regex.MatchString(DecodeOnce(target))
 	}
 
-	return rule.regex.MatchString(value(rule.Target, r))
+	return rule.regex.MatchString(target)
 }
 
 // Path returns r's path as the client sent it, before any decoding or
-// re-encoding, up to the first ?: what a path rule is matched against.
+// re-encoding, up to the first ?: what a path rule is matched against, as
+// it is and once percent-decoded.
 func Path(r *http.Request) string {
 	path, _, _ := strings.Cut(pathAndQuery(r), "?")
 
 	return path
 }
 
-// value returns what a rule with target, other than uri, is matched
+// value returns what a rule with target is matched against in r: the
-// against in r: the path, as Path gives it, and the query as the client
+// path, as Path gives it, the path and query, as pathAndQuery gives
-// sent it, before any decoding or re-encoding, after the first ?, and a
+// them, and the query as the client sent it, before any decoding or
-// header's values joined by ", ", as HTTP joins those of a header sent
+// re-encoding, after the first ?, and a header's values joined by ", ",
-// more than once.
+// as HTTP joins those of a header sent more than once.
 func value(target string, r *http.Request) string {
 	switch target {
-	case "path":
+	case pathTarget:
 		return Path(r)
+	case uriTarget:
+		return pathAndQuery(r)
 	case "query":
 		_, query, _ := strings.Cut(pathAndQuery(r), "?")
 
@@ -464,11 +474,11 @@ func header(r *http.Request, name string) string {
 	return strings.Join(r.Header.Values(name), ", ")
 }
 
-// decodeOnce returns s with each percent escape, such as %2e, replaced by
+// DecodeOnce returns s with each percent escape, such as %2e, replaced by
 // the byte it stands for. A % that is not followed by two hex digits is
 // left as it is, so that a malformed escape cannot keep the rest of s
 // from being decoded.
-func decodeOnce(s string) string {
+func DecodeOnce(s string) string {
 	var decoded strings.Builder
 
 	for i := 0; i < len(s); i++ {

internal/rules/rules_test.go

@@ -49,7 +49,9 @@ func TestEachTargetMatchesWhatItNames(t *testing.T) {
 		want   bool
 	}{
 		{"path as received", `path log ^/%2eenv$`, "/%2eenv", nil, true},
-		{"path not decoded", `path log ^/\.env$`, "/%2eenv", nil, false},
+		{"path decoded", `path log ^/\.env$`, "/%2Eenv", nil, true},
+		{"path decoded only once", `path log ^/\.env$`, "/%252eenv", nil, false},
+		{"path decoded without the query", `path log ^/a\?b$`, "/a%3fb?c", nil, true},
 		{"path without the query", `path log ^/a$`, "/a?b=c", nil, true},
 		{"query as received", `query log ^b=%2e$`, "/a?b=%2e", nil, true},
 		{"uri as received", `uri log ^/a\?b=%2e$`, "/a?b=%2e", nil, true},
@@ -418,7 +420,7 @@ func TestDefaultFileBansProbesAtTheSiteRootAlone(t *testing.T) {
 
 	// Probes sent by a browser, by the rule that refuses them.
 	for rule, targets := range map[string][]string{
-		"env-file":    {"/.env", "/.env.production", "/.ENV"},
+		"env-file":    {"/.env", "/.env.production", "/.ENV", "/%2eenv"},
 		"vcs-dir":     {"/.git/config", "/.git", "/.svn/entries"},
 		"secrets-dir": {"/.aws/credentials", "/.ssh/id_rsa"},
 		"secret-file": {"/.htpasswd", "/.DS_Store", "/.git-credentials"},

internal/waf/waf.go

@@ -1,12 +1,18 @@
 // Package waf runs the OWASP Core Rule Set 4.25.0, through Coraza, on the
-// method, the URL with its query and the headers of a request, with the
+// method, the URL with its query and the headers of a request, and on its
-// six changes smallwebwaf makes to it, as "Attack detection" under
+// body while SWWAF_WAF_BODY_LIMIT is set, with the six changes smallwebwaf
-// "Configuration surface" in SPEC.md describes them. It reads no request
+// makes to it, as "Attack detection" under "Configuration surface" in
-// body and no response.
+// SPEC.md describes them. It reads no response.
+//
+// smallwebwaf writes only to its state directory, so Coraza is built with
+// its no_fs_access tag, as the Dockerfile and script/build build it: of a
+// file in a multipart body, Coraza then counts the bytes instead of
+// writing them to the system's temporary directory.
 package waf
 
 import (
 	"fmt"
+	"io"
 	"net/http"
 	"net/netip"
 	"slices"
@@ -20,15 +26,16 @@ import (
 )
 
 // directives are the Core Rule Set as smallwebwaf runs it, with the
-// paranoia level for %d. Each rule smallwebwaf adds has an id from 900000
+// paranoia level for %d, and bodyDirectives for %s while
-// to 900999, the ids the Core Rule Set keeps for the rules that set it
+// SWWAF_WAF_BODY_LIMIT is set. Each rule smallwebwaf adds has an id from
-// up, which SWWAF_WAF_DISABLED_RULES refuses, so that no setting switches
+// 900000 to 900999, the ids the Core Rule Set keeps for the rules that set
-// one off. Coraza joins a line ending in \ to the next, without the spaces
+// it up, which SWWAF_WAF_DISABLED_RULES refuses, so that no setting
-// at the start of the next.
+// switches one off. Coraza joins a line ending in \ to the next, without
+// the spaces at the start of the next.
 const directives = `
 # The engine only detects. smallwebwaf compares the request's anomaly
 # score with SWWAF_WAF_ANOMALY_THRESHOLD itself, in block and detect mode
-# alike. It reads no body.
+# alike. It reads no body, unless bodyDirectives switch that on.
 SecRuleEngine DetectionOnly
 SecRequestBodyAccess Off
 SecResponseBodyAccess Off
@@ -44,19 +51,33 @@ SecAction "id:900200,phase:1,pass,nolog,\
     setvar:'tx.allowed_methods=GET HEAD POST OPTIONS PUT PATCH DELETE'"
 
 # The second: Expect and Content-Encoding are taken off the Core Rule Set's
-# list of the headers it refuses. Content-Encoding stays refused on a body
+# list of the headers it refuses. Content-Encoding goes back on it for a
-# the Core Rule Set reads, and it reads none.
+# body the Core Rule Set reads (900260 in bodyDirectives).
 SecAction "id:900250,phase:1,pass,nolog,\
     setvar:'tx.restricted_headers_basic=/proxy/ /lock-token/ /content-range/ \
     /if/ /x-http-method-override/ /x-http-method/ /x-method-override/ \
     /x-middleware-subrequest/'"
+%s
+# Coraza keeps the first 1000 query parameters of a request, and the first
+# 1000 fields of a form data or JSON body, and drops the rest, which no
+# rule then reads, so a request with more adds 5 to the score, as a rule
+# the Core Rule Set rates critical does. Coraza's recommended
+# configuration refuses such a request in its rules 200004 and 200005.
+# This rule runs once the body is read, and before the Core Rule Set adds
+# up the score in the same phase.
+SecArgumentsLimit 1000
+SecRule ARGUMENTS_LIMIT_REACHED "@eq 1" "id:900300,phase:2,pass,\
+    severity:'CRITICAL',setvar:'tx.inbound_anomaly_score_pl1=+5'"
 
 # The sixth: only the rules for requests are loaded, and no response is
 # inspected.
 Include @owasp_crs/REQUEST-*.conf
 
 # The third: redirect_uri is not checked for a URL naming an IP address or
-# localhost.
+# localhost. Coraza matches a parameter name here, and in the fourth,
+# without regard to case. ARGS holds the fields of a form data or multipart
+# body Coraza reads as well as the query parameters, so a field of one of
+# these names is left out too.
 SecRuleUpdateTargetById 931100 "!ARGS:redirect_uri"
 SecRuleUpdateTargetById 934110 "!ARGS:redirect_uri"
 
@@ -79,15 +100,47 @@ SecRuleUpdateTargetById 932260 "!ARGS:path|!ARGS:files|!ARGS:skip-to|\
 # Inspect.
 SecRuleUpdateTargetById 932340 "!REQUEST_HEADERS:Referer"
 SecRuleUpdateTargetById 944110 "!REQUEST_HEADERS:Referer"
+`
 
-# Coraza keeps the first 1000 query parameters of a request and drops the
+// bodyDirectives have the Core Rule Set read the part of a request body
-# rest, which no rule then reads, so a request with more adds 5 to the
+// Inspect gives it, which is at most one byte longer than the limit, up to
-# score, as a rule the Core Rule Set rates critical does. Coraza's
+// the limit, %d bytes, and read JSON and XML as Coraza's recommended
-# recommended configuration refuses such a request in its rule 200004.
+// configuration has it in its rules 200000, 200001 and 200006, with
-# This rule comes after the Core Rule Set's, which set the score to 0 in
+// text/json, and any application or text type ending in +xml or +json,
-# the same phase.
+// besides; form data and multipart Coraza knows by itself. %% stands for
-SecArgumentsLimit 1000
+// a % Coraza reads.
-SecRule ARGUMENTS_LIMIT_REACHED "@eq 1" "id:900300,phase:1,pass,\
+const bodyDirectives = `
+SecRequestBodyAccess On
+SecRequestBodyLimit %d
+SecRequestBodyLimitAction ProcessPartial
+
+SecRule REQUEST_HEADERS:Content-Type \
+    "@rx ^(?:application|text)/(?:[a-z0-9.-]+[+])?xml" \
+    "id:900410,phase:1,pass,nolog,t:none,t:lowercase,ctl:requestBodyProcessor=XML"
+SecRule REQUEST_HEADERS:Content-Type \
+    "@rx ^(?:application|text)/(?:[a-z0-9.-]+[+])?json" \
+    "id:900420,phase:1,pass,nolog,t:none,t:lowercase,ctl:requestBodyProcessor=JSON"
+
+# The rest of the second change: Content-Encoding is refused again on a
+# body of a kind the Core Rule Set reads, since a compressed body cannot be
+# inspected.
+SecRule REQBODY_PROCESSOR "@rx ^(?:URLENCODED|MULTIPART|JSON|XML)$" \
+    "id:900260,phase:1,pass,nolog,\
+    setvar:'tx.restricted_headers_basic=%%{tx.restricted_headers_basic} \
+    /content-encoding/'"
+
+# A body Coraza fails to parse (900440), and a multipart body that fails
+# its strict checks (900450), each add 5 to the score, as a rule the Core
+# Rule Set rates critical does: no rule reads what comes after the fault,
+# which the app may still read. Coraza's recommended configuration refuses
+# them in its rules 200002 and 200003. A multipart body the limit cuts
+# before the colon of a part's header line, or between the carriage return
+# and the line feed that end a part's header line or the empty line after
+# its headers, adds 5 too, since Coraza takes the line the limit cuts for a
+# malformed header. Coraza parses any form data body.
+SecRule REQBODY_ERROR "!@eq 0" "id:900440,phase:2,pass,severity:'CRITICAL',\
+    setvar:'tx.inbound_anomaly_score_pl1=+5'"
+SecRule MULTIPART_STRICT_ERROR "!@eq 0" "id:900450,phase:2,pass,\
     severity:'CRITICAL',setvar:'tx.inbound_anomaly_score_pl1=+5'"
 `
 
@@ -104,18 +157,28 @@ type Params struct {
 	// DisabledRules are the ids of the rules switched off
 	// (SWWAF_WAF_DISABLED_RULES).
 	DisabledRules []int
+	// BodyLimit is the most of a request body the Core Rule Set reads
+	// (SWWAF_WAF_BODY_LIMIT), 0 while it is off and it reads none.
+	BodyLimit int64
 }
 
 // CoreRuleSet is the Core Rule Set, ready to inspect requests. It is safe
 // for concurrent use.
 type CoreRuleSet struct {
-	waf coraza.WAF
+	waf       coraza.WAF
+	bodyLimit int64
 }
 
 // New returns the Core Rule Set with the six changes, at params'
-// paranoia level and without the rules it switches off.
+// paranoia level, without the rules it switches off, and reading request
+// bodies up to params' limit.
 func New(params Params) (*CoreRuleSet, error) {
-	text := fmt.Sprintf(directives, params.ParanoiaLevel)
+	body := ""
+	if params.BodyLimit > 0 {
+		body = fmt.Sprintf(bodyDirectives, params.BodyLimit)
+	}
+
+	text := fmt.Sprintf(directives, params.ParanoiaLevel, body)
 
 	if len(params.DisabledRules) > 0 {
 		ids := make([]string, len(params.DisabledRules))
@@ -133,7 +196,7 @@ func New(params Params) (*CoreRuleSet, error) {
 		return nil, fmt.Errorf("load the Core Rule Set: %w", err)
 	}
 
-	return &CoreRuleSet{waf: waf}, nil
+	return &CoreRuleSet{waf: waf, bodyLimit: params.BodyLimit}, nil
 }
 
 // Result is what the Core Rule Set found in a request.
@@ -147,10 +210,15 @@ type Result struct {
 
 // Inspect runs the Core Rule Set on r, a request from client: on its
 // method, its URL with the query, and its headers, the Cookie header
-// without the cookies in cookiesNotRead.
+// without the cookies in cookiesNotRead, and on body, r's body as the
-func (c *CoreRuleSet) Inspect(r *http.Request, client netip.Addr) Result {
+// caller has it, as readBody reads it. It returns what it found, what it
+// read of body, which the app is still to be sent, and the error that
+// ended the reading early, if one did.
+func (c *CoreRuleSet) Inspect(
+	r *http.Request, client netip.Addr, body io.Reader,
+) (Result, []byte, error) {
 	tx := c.waf.NewTransaction()
-	// With no body read, there is nothing whose closing can fail.
+	// Closing would remove the files Coraza wrote, and it writes none.
 	defer func() { _ = tx.Close() }()
 
 	tx.ProcessConnection(client.String(), 0, "", 0)
@@ -177,7 +245,11 @@ func (c *CoreRuleSet) Inspect(r *http.Request, client netip.Addr) Result {
 	}
 
 	tx.ProcessRequestHeaders()
-	// With no body read, this runs the rest of the rules, and cannot fail.
+
+	read, err := c.readBody(tx, body)
+
+	// This reads the body in memory and runs the rest of the rules, and
+	// cannot fail.
 	_, _ = tx.ProcessRequestBody()
 
 	var ids []int
@@ -191,7 +263,44 @@ func (c *CoreRuleSet) Inspect(r *http.Request, client netip.Addr) Result {
 		}
 	}
 
-	return Result{RuleIDs: ids, Score: score(tx)}
+	return Result{RuleIDs: ids, Score: score(tx)}, read, err
+}
+
+// readBody reads body, the body of the request in tx, which has run on
+// the request's headers, while SWWAF_WAF_BODY_LIMIT is set and the body is
+// of a kind the Core Rule Set reads: form data and multipart, of which it
+// reads the first c.bodyLimit bytes, and JSON and XML, which it reads only
+// when they are no longer than that, since they cannot be read in part.
+// readBody reads one byte past the limit, to tell which they are, gives
+// the Core Rule Set what it reads, and returns what it read and the error
+// that ended the reading early, if one did.
+func (c *CoreRuleSet) readBody(tx types.Transaction, body io.Reader) ([]byte, error) {
+	if c.bodyLimit == 0 {
+		return nil, nil
+	}
+
+	inPart := false
+	// How the body is read is a variable of the transaction, which only
+	// Coraza's interface for plugins reads.
+	state := tx.(plugintypes.TransactionState) //nolint:forcetypeassert // every one is
+
+	switch state.Variables().RequestBodyProcessor().Get() {
+	case "URLENCODED", "MULTIPART":
+		inPart = true
+	case "JSON", "XML":
+	default:
+		return nil, nil
+	}
+
+	read, err := io.ReadAll(io.LimitReader(body, c.bodyLimit+1))
+
+	if inPart || (err == nil && int64(len(read)) <= c.bodyLimit) {
+		// Coraza holds what it reads of the body in memory, up to the
+		// limit, so this cannot fail.
+		_, _, _ = tx.WriteRequestBody(read)
+	}
+
+	return read, err
 }
 
 // score returns the anomaly score the Core Rule Set added up in tx, a

internal/waf/waf_test.go

@@ -4,7 +4,10 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"net/netip"
+	"net/url"
+	"path/filepath"
 	"reflect"
+	"strconv"
 	"strings"
 	"testing"
 
@@ -48,12 +51,28 @@ func get(target string, headers ...string) request {
 func inspect(t *testing.T, crs *waf.CoreRuleSet, r request) waf.Result {
 	t.Helper()
 
+	result, _ := inspectBody(t, crs, r, "")
+
+	return result
+}
+
+// inspectBody is inspect for r with body, which is announced with its
+// Content-Length unless it is "", and returns what crs read of body too.
+func inspectBody(
+	t *testing.T, crs *waf.CoreRuleSet, r request, body string,
+) (waf.Result, string) {
+	t.Helper()
+
 	req := httptest.NewRequestWithContext(t.Context(), r.method,
-		"http://git.example"+r.target, http.NoBody)
+		"http://git.example"+r.target, strings.NewReader(body))
 	req.Header.Set("User-Agent", "Mozilla/5.0 (X11; Linux x86_64; rv:131.0) "+
 		"Gecko/20100101 Firefox/131.0")
 	req.Header.Set("Accept", "text/html")
 
+	if body != "" {
+		req.Header.Set("Content-Length", strconv.Itoa(len(body)))
+	}
+
 	for _, header := range r.headers {
 		// Go's server keeps Host and Transfer-Encoding out of the headers.
 		name, value, _ := strings.Cut(header, ": ")
@@ -67,7 +86,12 @@ func inspect(t *testing.T, crs *waf.CoreRuleSet, r request) waf.Result {
 		}
 	}
 
-	return crs.Inspect(req, netip.MustParseAddr("203.0.113.9"))
+	result, read, err := crs.Inspect(req, netip.MustParseAddr("203.0.113.9"), req.Body)
+	if err != nil {
+		t.Fatalf("read the body: %v", err)
+	}
+
+	return result, string(read)
 }
 
 // wantResult checks what crs finds in r.
@@ -139,7 +163,9 @@ func TestExpectAndContentEncodingAreAllowed(t *testing.T) {
 		request{http.MethodPost, push, []string{pushType, length, "Expect: 100-continue"}},
 		waf.Result{})
 	wantResult(t, crs,
-		request{http.MethodPost, fetch, []string{fetchType, length, "Content-Encoding: gzip"}},
+		request{http.MethodPost, fetch, []string{
+			fetchType, length, "Content-Encoding: gzip",
+		}},
 		waf.Result{})
 
 	// Every other header on the Core Rule Set's list stays refused.
@@ -172,7 +198,7 @@ func TestTransferEncodingIsRead(t *testing.T) {
 		waf.Result{})
 }
 
-func TestMoreQueryParametersThanCorazaKeepsIsAMatch(t *testing.T) {
+func TestMoreParametersThanCorazaKeepsIsAMatch(t *testing.T) {
 	t.Parallel()
 
 	const attack = "id=1'%20OR%20'1'='1"
@@ -183,6 +209,23 @@ func TestMoreQueryParametersThanCorazaKeepsIsAMatch(t *testing.T) {
 	// after it is not, but the request is a match all the same.
 	wantResult(t, crs, get("/?"+strings.Repeat("a=1&", 999)+attack), matched(942100))
 	wantResult(t, crs, get("/?"+strings.Repeat("a=1&", 1000)+attack), matched(900300))
+
+	// So it is with the fields of a form data or JSON body.
+	crs = readingBodies(t)
+
+	for _, tc := range []struct{ header, body string }{
+		{formData, strings.Repeat("a=1&", 999) + attack},
+		{jsonBody, `{"a":[` + strings.Repeat("1,", 998) + `1],"id":"` + injection + `"}`},
+	} {
+		wantBody(t, crs, post(tc.header), tc.body, matched(942100), tc.body)
+	}
+
+	for _, tc := range []struct{ header, body string }{
+		{formData, strings.Repeat("a=1&", 1000) + attack},
+		{jsonBody, `{"a":[` + strings.Repeat("1,", 999) + `1],"id":"` + injection + `"}`},
+	} {
+		wantBody(t, crs, post(tc.header), tc.body, matched(900300), tc.body)
+	}
 }
 
 func TestRedirectURIMayNameALocalAddress(t *testing.T) {
@@ -231,6 +274,16 @@ func TestParametersGiteaSendsNamesInSkipTheListsOfFilesPathsAndCommands(t *testi
 	wantResult(t, crs, get("/?path=1'%20OR%20'1'='1"), matched(942100))
 }
 
+func TestParameterNamesAreMatchedWithoutRegardToCase(t *testing.T) {
+	t.Parallel()
+
+	crs := atDefaults(t)
+
+	wantChange(t, crs, get("/?Path=.gitignore"), get("/?q=.gitignore"), matched(930120))
+	wantChange(t, crs, get("/?REDIRECT_URI=http://127.0.0.1:52341/"),
+		get("/?next=http://127.0.0.1:52341/"), matched(931100, 934110))
+}
+
 func TestCookiesGiteaFlashAndRedirectToAreNotRead(t *testing.T) {
 	t.Parallel()
 
@@ -308,3 +361,328 @@ func TestEachDisabledRuleIsSwitchedOff(t *testing.T) {
 		waf.Result{RuleIDs: []int{911100, 920350}, Score: 8})
 	wantResult(t, newCoreRuleSet(t, 1, 920350, 911100), r, waf.Result{})
 }
+
+// bodyLimit is SWWAF_WAF_BODY_LIMIT in the tests that read bodies.
+const bodyLimit = 8 << 10
+
+// The Content-Type headers of the kinds of body the Core Rule Set reads.
+const (
+	formData  = "Content-Type: application/x-www-form-urlencoded"
+	multipart = "Content-Type: multipart/form-data; boundary=b"
+	jsonBody  = "Content-Type: application/json"
+	xmlBody   = "Content-Type: application/xml"
+)
+
+// injection is an SQL injection, which rule 942100 matches.
+const injection = "1' OR '1'='1"
+
+// readingBodies returns the Core Rule Set as smallwebwaf runs it by
+// default, but reading bodies up to bodyLimit.
+func readingBodies(t *testing.T) *waf.CoreRuleSet {
+	t.Helper()
+
+	crs, err := waf.New(waf.Params{
+		ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
+	})
+	if err != nil {
+		t.Fatalf("load the Core Rule Set: %v", err)
+	}
+
+	return crs
+}
+
+// post is a POST request for / with a body of the type contentType, a
+// Content-Type header, gives, and headers besides.
+func post(contentType string, headers ...string) request {
+	return request{http.MethodPost, "/", append([]string{contentType}, headers...)}
+}
+
+// field is a part of a multipart body: the field name, holding value.
+func field(name, value string) string {
+	return "--b\r\nContent-Disposition: form-data; name=\"" + name + "\"\r\n\r\n" +
+		value + "\r\n"
+}
+
+// end ends a multipart body.
+const end = "--b--\r\n"
+
+// padded returns head and tail with as many a's between them as make n
+// bytes in all.
+func padded(head, tail string, n int) string {
+	return head + strings.Repeat("a", n-len(head)-len(tail)) + tail
+}
+
+// wantBody checks what crs finds in r with body, and that what it read of
+// body is read.
+func wantBody(
+	t *testing.T, crs *waf.CoreRuleSet, r request, body string, want waf.Result,
+	read string,
+) {
+	t.Helper()
+
+	got, gotRead := inspectBody(t, crs, r, body)
+	if !reflect.DeepEqual(got, want) || gotRead != read {
+		t.Errorf("%q with a body of %d bytes, %.40q: %+v, reading %d bytes, "+
+			"want %+v, reading %d", r.headers, len(body), body, got, len(gotRead),
+			want, len(read))
+	}
+}
+
+func TestBodiesAreReadOnlyWhileBodyLimitIsSet(t *testing.T) {
+	t.Parallel()
+
+	off, on := atDefaults(t), readingBodies(t)
+
+	for _, tc := range []struct{ header, body string }{
+		{formData, "q=" + url.QueryEscape(injection)},
+		{multipart, field("q", injection) + end},
+		{jsonBody, `{"q":"` + injection + `"}`},
+		{xmlBody, "<q>" + injection + "</q>"},
+	} {
+		wantBody(t, off, post(tc.header), tc.body, waf.Result{}, "")
+		wantBody(t, on, post(tc.header), tc.body, matched(942100), tc.body)
+	}
+}
+
+func TestFormDataAndMultipartAreReadUpToTheLimit(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+	pad := strings.Repeat("a", bodyLimit)
+
+	for _, tc := range []struct{ header, attackFirst, attackLast string }{
+		{
+			formData, "q=" + url.QueryEscape(injection) + "&pad=" + pad,
+			"pad=" + pad + "&q=" + url.QueryEscape(injection),
+		},
+		{
+			multipart, field("q", injection) + field("pad", pad) + end,
+			field("pad", pad) + field("q", injection) + end,
+		},
+	} {
+		wantBody(t, crs, post(tc.header), tc.attackFirst, matched(942100),
+			tc.attackFirst[:bodyLimit+1])
+		wantBody(t, crs, post(tc.header), tc.attackLast, waf.Result{},
+			tc.attackLast[:bodyLimit+1])
+	}
+
+	// To the byte: a system file's path is found when it ends at the limit,
+	// and not when its last letter is past it, which is still read.
+	atLimit := padded("pad=", "&q=/etc/passwd", bodyLimit)
+	wantBody(t, crs, post(formData), atLimit, matched(930120, 932160), atLimit)
+
+	pastLimit := padded("pad=", "&q=/etc/passwd", bodyLimit+1)
+	wantBody(t, crs, post(formData), pastLimit, waf.Result{}, pastLimit)
+}
+
+func TestJSONAndXMLAreReadOnlyWhenNoLargerThanTheLimit(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+
+	for _, tc := range []struct{ header, head, tail string }{
+		{jsonBody, `{"q":"` + injection + `","pad":"`, `"}`},
+		{xmlBody, "<r><q>" + injection + "</q><pad>", "</pad></r>"},
+	} {
+		fits := padded(tc.head, tc.tail, bodyLimit)
+		wantBody(t, crs, post(tc.header), fits, matched(942100), fits)
+
+		larger := padded(tc.head, tc.tail, bodyLimit+1)
+		wantBody(t, crs, post(tc.header), larger, waf.Result{}, larger)
+	}
+}
+
+func TestOtherBodiesAreNotRead(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+
+	// Read as form data, which the Core Rule Set does with a body of a type
+	// it does not know, this would be an SQL injection.
+	body := "q=" + url.QueryEscape(injection)
+
+	for _, header := range []string{
+		"Content-Type: application/octet-stream",
+		"Content-Type: text/plain",
+		"Content-Type: application/x-git-receive-pack-request",
+	} {
+		wantBody(t, crs, post(header), body, waf.Result{}, "")
+	}
+}
+
+func TestContentEncodingIsRefusedOnTheKindsOfBodyTheCoreRuleSetReads(t *testing.T) {
+	t.Parallel()
+
+	const gzip = "Content-Encoding: gzip"
+
+	crs := readingBodies(t)
+
+	for _, tc := range []struct{ header, body string }{
+		{formData, "a=1"},
+		{multipart, field("a", "1") + end},
+		{jsonBody, `{"a":1}`},
+		{xmlBody, "<a>1</a>"},
+	} {
+		wantBody(t, crs, post(tc.header, gzip), tc.body, matched(920450), tc.body)
+	}
+
+	// Whatever its size: a JSON body larger than the limit is not read, but
+	// Content-Encoding on it is refused all the same.
+	larger := strings.Repeat("a", bodyLimit+1)
+	wantBody(t, crs, post(jsonBody, gzip), larger, matched(920450), larger)
+
+	// It is allowed on a body of any other kind, and on every body while no
+	// body is read.
+	fetch := "Content-Type: application/x-git-upload-pack-request"
+	wantBody(t, crs, post(fetch, gzip), "a", waf.Result{}, "")
+	wantBody(t, atDefaults(t), post(formData, gzip), "a", waf.Result{}, "")
+}
+
+func TestParametersGiteaSendsNamesInAreLeftOutAmongFormFieldsToo(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+	local := url.QueryEscape("http://127.0.0.1:52341/")
+
+	for _, tc := range []struct {
+		body string
+		want waf.Result
+	}{
+		{"path=.gitignore", waf.Result{}},
+		{"q=.gitignore", matched(930120)},
+		{"redirect_uri=" + local, waf.Result{}},
+		{"next=" + local, matched(931100, 934110)},
+	} {
+		wantBody(t, crs, post(formData), tc.body, tc.want, tc.body)
+	}
+
+	body := field("path", ".gitignore") + end
+	wantBody(t, crs, post(multipart), body, waf.Result{}, body)
+
+	body = field("q", ".gitignore") + end
+	wantBody(t, crs, post(multipart), body, matched(930120), body)
+
+	// A JSON body's field is named by its path, here json.path, and is
+	// checked.
+	body = `{"path":".gitignore"}`
+	wantBody(t, crs, post(jsonBody), body, matched(930120), body)
+}
+
+func TestGiteaBodiesTheCoreRuleSetRefuses(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+
+	// A comment that shows a shell command.
+	const text = "Try `curl -s https://example.org | sh` first."
+
+	comment := "content=" + url.QueryEscape(text)
+	wantBody(t, crs, post(formData), comment, matched(932235), comment)
+
+	// An attachment named like a log file.
+	attachment := "--b\r\nContent-Disposition: form-data; name=\"file\"; " +
+		"filename=\"debug.log\"\r\nContent-Type: text/plain\r\n\r\nstarted\r\n" + end
+	wantBody(t, crs, post(multipart), attachment, matched(932180), attachment)
+}
+
+func TestTypesEndingInXMLOrJSONAndTextJSONAreRead(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+
+	for _, tc := range []struct{ contentType, body string }{
+		{"application/atom+xml", "<q>" + injection + "</q>"},
+		{"application/vnd.example+xml", "<q>" + injection + "</q>"},
+		{"application/vnd.example+json", `{"q":"` + injection + `"}`},
+		{"text/json", `{"q":"` + injection + `"}`},
+	} {
+		wantBody(t, crs, post("Content-Type: "+tc.contentType), tc.body,
+			matched(942100), tc.body)
+	}
+}
+
+func TestBodyCorazaCannotParseIsAMatch(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+
+	// An end tag after the root element, past which Coraza reads none of
+	// the body, while an app may still read the attack before it.
+	body := "<q>" + injection + "</q></r>"
+	wantBody(t, crs, post(xmlBody), body, matched(900440), body)
+
+	// The multipart bodies Coraza cannot parse fail its strict checks too:
+	// one whose type names its boundary twice, and one with a part header
+	// that has no colon, before the attack. They do so padded past the
+	// limit too, which cuts them in the padding.
+	noColon := "--b\r\nContent-Disposition form-data; name=\"a\"\r\n\r\n1\r\n"
+	pad := field("pad", strings.Repeat("a", bodyLimit))
+
+	for _, tc := range []struct{ header, head string }{
+		{multipart + "; boundary=c", ""},
+		{multipart, noColon},
+	} {
+		body = tc.head + field("q", injection) + end
+		wantBody(t, crs, post(tc.header), body, matched(900440, 900450), body)
+
+		body = tc.head + field("q", injection) + pad + end
+		wantBody(t, crs, post(tc.header), body, matched(900440, 900450),
+			body[:bodyLimit+1])
+	}
+}
+
+func TestMultipartBodyCutBeforeAPartHeadersColonIsAMatch(t *testing.T) {
+	t.Parallel()
+
+	// The limit falls in the middle of the name of the second part's
+	// header, which Coraza, reading up to the limit, cannot tell from a
+	// header without a colon.
+	cut := "--b\r\nContent-Di"
+	first := field("pad", strings.Repeat("a", bodyLimit-len(field("pad", ""))-len(cut)))
+	body := first + cut + "sposition: form-data; name=\"q\"\r\n\r\n1\r\n" + end
+
+	wantBody(t, readingBodies(t), post(multipart), body, matched(900440, 900450),
+		body[:bodyLimit+1])
+}
+
+func TestMultipartBodyCutBeforeALineFeedInAPartsHeadersIsAMatch(t *testing.T) {
+	t.Parallel()
+
+	crs := readingBodies(t)
+	headerLine := "--b\r\nContent-Disposition: form-data; name=\"q\"\r"
+
+	// The limit falls between the carriage return and the line feed that
+	// end the second part's header line, and then between those that end
+	// the empty line after it. Coraza, reading up to the limit, takes the
+	// line ending in a lone carriage return for a malformed header.
+	for _, cut := range []int{len(headerLine), len(headerLine + "\n\r")} {
+		first := field("pad", strings.Repeat("a", bodyLimit-len(field("pad", ""))-cut))
+		body := first + field("q", "1") + end
+
+		wantBody(t, crs, post(multipart), body, matched(900440, 900450),
+			body[:bodyLimit+1])
+	}
+}
+
+// TestCorazaWritesNoFile is not parallel, since it sets TMPDIR, the
+// system's temporary directory, for the whole test process.
+func TestCorazaWritesNoFile(t *testing.T) {
+	// The system's temporary directory is one that does not exist, so that
+	// Coraza could write nothing there: built without no_fs_access, it
+	// refuses to load, and could not write a file of a multipart body.
+	t.Setenv("TMPDIR", filepath.Join(t.TempDir(), "missing"))
+
+	body := "--b\r\nContent-Disposition: form-data; name=\"file\"; " +
+		"filename=\"notes.txt\"\r\nContent-Type: text/plain\r\n\r\n" +
+		strings.Repeat("a", 1000) + "\r\n" + field("q", injection) + end
+	wantBody(t, readingBodies(t), post(multipart), body, matched(942100), body)
+}
+
+func TestBodyLimitOf1GLoads(t *testing.T) {
+	t.Parallel()
+
+	_, err := waf.New(waf.Params{ParanoiaLevel: 1, BodyLimit: 1 << 30})
+	if err != nil {
+		t.Errorf("load the Core Rule Set reading bodies up to 1G: %v", err)
+	}
+}

script/build

@@ -1,7 +1,7 @@
 #!/bin/sh
 # script/build: build bin/smallwebwaf on the host, with Go installed, for
 # working on the code by hand. The version it reports comes from git, as
-# in script/docker.
+# in script/docker, and the no_fs_access tag is the Dockerfile's.
 set -eu
 
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -11,7 +11,7 @@ main() {
     cd "$ROOT"
     version="$(git describe --tags --always --dirty 2>/dev/null || true)"
     [ -n "$version" ] || version="unknown"
-    go build -trimpath -ldflags "-X main.Version=$version" \
+    go build -tags no_fs_access -trimpath -ldflags "-X main.Version=$version" \
         -o bin/smallwebwaf ./cmd/smallwebwaf
 }