add cl-parser.c

4 files changed, 184 insertions, 3 deletions

diff --git a/Makefile b/Makefile
index 0ab3d8f..0de2951 100644
--- a/Makefile
+++ b/Makefile
@@ -1,6 +1,6 @@
 CC ?= cc
 CFLAGS=-std=c99 -W -Wall -pedantic -O2 -Iinclude -fPIC
-OBJS=fhp.o hash.o error.o token.o env.o te-parser.o header-value-parser.o
+OBJS=fhp.o hash.o error.o token.o env.o te-parser.o cl-parser.o header-value-parser.o
 SONAME=libfhp.so
 LIB=libfhp.so
 STATIC_LIB=libfhp.a
diff --git a/cl-parser.c b/cl-parser.c
new file mode 100644
index 0000000..83863c5
--- /dev/null
+++ b/cl-parser.c
@@ -0,0 +1,150 @@
+#include "internal.h"
+
+//
+// content length functions
+//
+
+static const fhp_cl_parser_t FHP_DEFAULT_CL_PARSER = {
+  .state = FHP_CL_STATE_INIT,
+  .val = 0,
+  .err = FHP_OK,
+};
+
+fhp_err_t
+fhp_cl_parser_init(fhp_cl_parser_t * const p) {
+  // init parser
+  *p = FHP_DEFAULT_CL_PARSER;
+
+  // return succes
+  return FHP_OK;
+}
+
+static fhp_err_t
+fhp_cl_parser_push_byte(
+  fhp_cl_parser_t * const p,
+  uint8_t byte
+) {
+retry:
+  switch (p->state) {
+  case FHP_CL_STATE_INIT:
+    switch (byte) {
+    CASE_OWS_CHARS
+      // ignore
+      break;
+    CASE_DIGIT_CHARS
+      // set state
+      p->state = FHP_CL_STATE_DATA;
+      goto retry;
+
+      break;
+    default:
+      // invalid character before content-length
+      return FHP_ERR_INVALID_CHAR_BEFORE_CL;
+    }
+
+    break;
+  case FHP_CL_STATE_DATA:
+    switch (byte) {
+    CASE_OWS_CHARS
+      // set state
+      p->state = FHP_CL_STATE_TRAILING_SPACES;
+      goto retry;
+
+      break;
+    CASE_DIGIT_CHARS
+      {
+        // cache old value
+        uint64_t old_val = p->val;
+
+        // update value
+        p->val = p->val * 10 + (byte - '0');
+
+        // check for overflow
+        if (p->val < old_val)
+          return FHP_ERR_CL_OVERFLOW;
+      }
+
+      break;
+    default:
+      // invalid character in content-length
+      return FHP_ERR_INVALID_CHAR_IN_CL;
+    }
+
+    break;
+  case FHP_CL_STATE_TRAILING_SPACES:
+    switch (byte) {
+    CASE_OWS_CHARS
+      // ignore trailing spaces
+
+      break;
+    default:
+      // invalid character after content-length
+      return FHP_ERR_INVALID_CHAR_AFTER_CL;
+    }
+
+    break;
+  case FHP_CL_STATE_ERROR:
+    // return last error
+    return p->err;
+
+    break;
+  case FHP_CL_STATE_DONE:
+    // shouldn't be reached
+    return FHP_ERR_CL_PARSER_DONE;
+
+    break;
+  default:
+    // never reached
+    return FHP_ERR_BAD_CL_STATE;
+  }
+
+  // return succes
+  return FHP_OK;
+}
+
+fhp_err_t
+fhp_cl_parser_push(
+  fhp_cl_parser_t *p,
+  uint8_t * const buf,
+  size_t len
+) {
+  for (size_t i = 0; i < len; i++) {
+    // push byte
+    fhp_err_t err = fhp_cl_parser_push_byte(p, buf[i]);
+
+    // check error
+    if (err != FHP_OK) {
+      // cache error
+      p->err = err;
+
+      // set state
+      p->state = FHP_CL_STATE_ERROR;
+
+      // return error
+      return err;
+    }
+  }
+
+  // return succes
+  return FHP_OK;
+}
+
+fhp_err_t
+fhp_cl_parser_done(
+  fhp_cl_parser_t * const p,
+  uint64_t * const ret_val
+) {
+  // check for error
+  if (p->err != FHP_OK)
+    return p->err;
+
+  // set state
+  p->state = FHP_TE_STATE_DONE;
+
+  // save number of tes
+  if (ret_val)
+    *ret_val = p->val;
+
+  // return success
+  return FHP_OK;
+}
diff --git a/error.c b/error.c
index 43ec598..9850de8 100644
--- a/error.c
+++ b/error.c
@@ -22,10 +22,16 @@ fhp_errors[] = {
   "too many transfer encodings",
   "invalid character in transfer encoding name",
   "invalid character in transfer encoding",
-  "bad transfer encoding state",
-  "transfer encoding parser already done",
+  "bad transfer-encoding parser state",
+  "transfer-encoding parser already done",
   "NULL transfer encoding hash output buffer",
   "transfer encoding hash output buffer is too small",
+  "invalid character before content-length header value",
+  "invalid character in content-length header value",
+  "invalid character after content-length header value",
+  "content-length parser already done",
+  "bad content-length parser state",
+  "overflow in content-length header value",
 };
 
 fhp_err_t
diff --git a/include/fhp/fhp.h b/include/fhp/fhp.h
index 46efb44..1e32a0d 100644
--- a/include/fhp/fhp.h
+++ b/include/fhp/fhp.h
@@ -41,6 +41,12 @@ typedef enum {
   FHP_ERR_TE_PARSER_DONE,
   FHP_ERR_NULL_TE_BUF,
   FHP_ERR_SMALL_TE_BUF,
+  FHP_ERR_INVALID_CHAR_BEFORE_CL,
+  FHP_ERR_INVALID_CHAR_IN_CL,
+  FHP_ERR_INVALID_CHAR_AFTER_CL,
+  FHP_ERR_CL_PARSER_DONE,
+  FHP_ERR_BAD_CL_STATE,
+  FHP_ERR_CL_OVERFLOW,
   FHP_ERR_LAST
 } fhp_err_t;
 
@@ -114,6 +120,25 @@ void fhp_env_init(fhp_env_t * const env);
 fhp_env_t *fhp_get_default_env(void);
 
 //
+// content length parser functions
+//
+
+typedef enum {
+  FHP_CL_STATE_INIT,
+  FHP_CL_STATE_DATA,
+  FHP_CL_STATE_TRAILING_SPACES,
+  FHP_CL_STATE_ERROR,
+  FHP_CL_STATE_DONE,
+  FHP_CL_STATE_LAST
+} fhp_cl_state_t;
+
+typedef struct {
+  fhp_cl_state_t state;
+  fhp_err_t err;
+  uint64_t val;
+} fhp_cl_parser_t;
+
+//
 // transfer encoding parser functions
 //