From 3f65a4ed2ee00e45072bf69ad6ed272d99096263 Mon Sep 17 00:00:00 2001
From: Kelvin Sherlock <ksherlock@gmail.com>
Date: Sun, 29 Apr 2012 22:46:22 -0400
Subject: [PATCH] http parsing

---
 htest.c      |  56 ++++++++++++++++++
 http.utils.c | 156 +++++++++++++++++++++++++++++++++++++++++++++++++++
 http.utils.h |  11 ++++
 3 files changed, 223 insertions(+)
 create mode 100644 htest.c
 create mode 100644 http.utils.c
 create mode 100644 http.utils.h

diff --git a/htest.c b/htest.c
new file mode 100644
index 0000000..0635398
--- /dev/null
+++ b/htest.c
@@ -0,0 +1,56 @@
+#include <stdio.h>
+#include <string.h>
+
+#include "http.utils.h"
+
+int main(int argc, char **argv)
+{
+
+    int version;
+    int status;
+    int ok;
+    char *cp;
+    
+    URLRange k, v;
+    
+    cp = "HTTP/1.0 200 OK";
+    ok = parseStatusLine(cp, strlen(cp), &version, &status);
+    if (!ok || version != 0x0100 || status != 200)
+    {
+        fprintf(stderr, "%s: %x %d\n", cp, version, status);
+    }
+
+    cp = "HTTP/1.1 404 bleh";
+    ok = parseStatusLine(cp, strlen(cp), &version, &status);
+    if (!ok || version != 0x0101 || status != 404)
+    {
+        fprintf(stderr, "%s: %x %d\n", cp, version, status);
+    }
+
+
+    cp = "http/1.10 200 asdasd as dsad asd";
+    ok = parseStatusLine(cp, strlen(cp), &version, &status);
+    if (!ok || version != 0x010a || status != 200)
+    {
+        fprintf(stderr, "%s: %x %d\n", cp, version, status);
+    }
+
+
+    cp = "key: value";
+    ok = parseHeaderLine(cp, strlen(cp), &k, &v);
+    if (!ok 
+        || k.location != 0 
+        || k.length != 3 
+        || v.location != 5 
+        || v.length !=  5)
+    {
+        fprintf(stderr, "%s: %.*s, %.*s\n", cp, 
+            k.length, cp + k.location,
+            v.length, cp + v.location);
+    }
+
+
+    return 0;
+}
+
+
diff --git a/http.utils.c b/http.utils.c
new file mode 100644
index 0000000..816de8f
--- /dev/null
+++ b/http.utils.c
@@ -0,0 +1,156 @@
+#ifdef __ORCAC__
+#pragma optimize 79
+#pragma noroot
+#endif
+
+#include <ctype.h>
+#include "url.h"
+
+
+int parseHeaderLine(const char *cp, unsigned length, URLRange *key, URLRange *value)
+{
+    unsigned i, l;
+    
+    key->location = 0;
+    key->length = 0;
+    value->location = 0;
+    value->length = 0;
+    
+    // trim any trailing whitespace.
+    
+    while (length)
+    {
+        if (isspace(cp[length - 1])) 
+            --length;
+        else break;
+    }
+    if (!length) return 0;
+    
+    /* format:
+     * key: value
+     * /^([^:]+):\s+(.*)$/
+     * -or-
+     *  value [continuation of previous line]
+     * /^\s+(.*)$/
+     */
+     
+    for (i = 0; i < length; ++i)
+    {
+      if (cp[i] == ':') break;
+    }
+    
+    if (i == length)
+    {
+      // try as value only
+      i = 0;
+    }
+    else
+    {
+      key->length = i;
+      i = i + 1;
+    }
+    
+    // now gobble up all the whitespace...
+    
+    for ( ; i < length; ++i)
+    {
+      if (!isspace(cp[i])) break;
+    }
+    
+    // no value? no problem!
+    if (i == length) return 1;
+    
+    value->location = i;
+    value->length = length - i;
+    
+    return 1;
+}
+
+
+int parseStatusLine(const char *cp, unsigned length, int *version, int *status)
+{
+    /*
+     * HTTP/1.1 200 OK etc.
+     *
+     */
+
+    unsigned short *wp;
+    int i;
+    char c;
+    int x;
+    
+    *version = 0;
+    *status = 0;
+    
+    wp = (unsigned short *)cp;
+
+    
+    // HTTP/
+    if (length <= 5) return 0;
+    if ((wp[0] | 0x2020) != 0x7468) return 0; // 'ht'
+    if ((wp[1] | 0x2020) != 0x7074) return 0; // 'tp'
+    if (cp[4] != '/') return 0;
+
+    // version string. 
+    // \d+ . \d+   
+    i = 5;
+    
+    c = cp[i];
+    if (!isdigit(c)) return 0;
+    x = c - '0';
+
+    for (i = i + 1; i < length; ++i)
+    {
+        c = cp[i];
+        if (!isdigit(c)) break;
+        x = (x << 1) + (x << 3) + (c - '0');
+    }
+    
+    *version = x << 8;
+    
+    if (i == length) return 0;
+    if (cp[i++] != '.') return 0;
+    
+    c = cp[i];
+    if (!isdigit(c)) return 0;
+    x = c - '0';
+
+    for (i = i + 1; i < length; ++i)
+    {
+        c = cp[i];
+        if (!isdigit(c)) break;
+        x = (x << 1) + (x << 3) + (c - '0');
+    }
+    
+    *version |= x;
+
+    // 1+ space
+    if (i == length) return 0;
+    c = cp[i];
+    if (!isspace(c)) return 0;
+
+    for (i = i + 1; i < length; ++i)
+    {
+        c = cp[i];
+        if (!isspace(c)) break;
+    }
+    
+    if (i == length) return 0;
+
+    c = cp[i];
+    if (!isdigit(c)) return 0;
+    x = c - '0';
+
+    for (i = i + 1; i < length; ++i)
+    {
+        c = cp[i];
+        if (!isdigit(c)) break;
+        x = (x << 1) + (x << 3) + (c - '0');
+    }
+    
+    *status = x;
+    
+    // rest of status line unimportant.
+    
+    return 1;
+}
diff --git a/http.utils.h b/http.utils.h
new file mode 100644
index 0000000..1a00b5d
--- /dev/null
+++ b/http.utils.h
@@ -0,0 +1,11 @@
+#ifndef __HTTP_UTILS_H__
+#define __HTTP_UTILS_H__
+
+#include "url.h"
+
+
+int parseHeaderLine(const char *line, unsigned length, URLRange *key, URLRange *value);
+
+int parseStatusLine(const char *cp, unsigned length, int *version, int *status);
+
+#endif
\ No newline at end of file