httpd: replace shell-based dir indexer cgi example with C-based one.
authorDenis Vlasenko <vda.linux@googlemail.com>
Wed, 22 Aug 2007 10:38:44 +0000 (10:38 -0000)
committerDenis Vlasenko <vda.linux@googlemail.com>
Wed, 22 Aug 2007 10:38:44 +0000 (10:38 -0000)
networking/httpd_index_cgi_example [deleted file]
networking/httpd_indexcgi.c [new file with mode: 0644]

diff --git a/networking/httpd_index_cgi_example b/networking/httpd_index_cgi_example
deleted file mode 100644 (file)
index 9c8e022..0000000
+++ /dev/null
@@ -1,50 +0,0 @@
-#!/bin/sh
-# This CGI creates directory index.
-# Put it into cgi-bin/index.cgi and chmod 0755.
-#
-# Problems:
-# * Unsafe wrt weird filenames with <>"'& etc...
-# * Not efficient: calls stat (program, not syscall) for each file
-# * Probably requires bash
-#
-# If you want speed and safety, you need to code it in C
-
-# Must start with '/'
-test "${QUERY_STRING:0:1}" = "/" || exit 1
-# /../ is not allowed
-test "${QUERY_STRING%/../*}" = "$QUERY_STRING" || exit 1
-test "${QUERY_STRING%/..}" = "$QUERY_STRING" || exit 1
-
-# Outta cgi-bin...
-cd .. 2>/dev/null || exit 1
-# Strip leading '/', go to target dir
-cd "${QUERY_STRING:1}" 2>/dev/null || exit 1
-
-f=`dirname "$QUERY_STRING"`
-test "$f" = "/" && f=""
-
-printf "%s" \
-$'HTTP/1.0 200 OK\r\n'\
-$'Content-type: text/html\r\n\r\n'\
-"<html><head><title>Index of $QUERY_STRING</title></head>"$'\r\n'\
-"<body><h1>Index of $QUERY_STRING</h1><pre>"$'\r\n'\
-$'<table width=100%>\r\n'\
-$'<col><col><col width=0*>\r\n'\
-$'<tr><th>Name<th align=right>Last modified<th align=right>Size\r\n'\
-\
-"<tr><td><a href='$f/'>..</a><td><td>"$'\r\n'
-
-IFS='#'
-for f in *; do
-    # Guard against empty dirs...
-    test -e "$f" && \
-    stat -c "%F#%s#%z" "$f" | {
-       read type size cdt junk
-       dir=''
-       test "$type" = "directory" && dir='/'
-       cdt="${cdt//.*}" # no fractional seconds
-       cdt="${cdt// /&nbsp;}" # prevent wrapping around space
-       printf "%s" "<tr><td><a href='$f$dir'>$f</a><td align=right>$cdt<td align=right>$size"$'\r\n'
-    }
-done
-printf "</table></pre><hr></body></html>"$'\r\n'
diff --git a/networking/httpd_indexcgi.c b/networking/httpd_indexcgi.c
new file mode 100644 (file)
index 0000000..b7e466c
--- /dev/null
@@ -0,0 +1,331 @@
+/*
+ * Copyright (c) 2007 Denys Vlasenko <vda.linux@googlemail.com>
+ *
+ * Licensed under GPLv2, see file LICENSE in this tarball for details.
+ */
+
+/*
+ * This program is a CGI application. It creates directory index page.
+ * Put it into cgi-bin/index.cgi and chmod 0755.
+ */
+
+/* Build a-la
+i486-linux-uclibc-gcc \
+-static -static-libgcc \
+-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 \
+-Wall -Wshadow -Wwrite-strings -Wundef -Wstrict-prototypes -Werror \
+-Wold-style-definition -Wdeclaration-after-statement -Wno-pointer-sign \
+-Wmissing-prototypes -Wmissing-declarations \
+-Os -fno-builtin-strlen -finline-limit=0 -fomit-frame-pointer \
+-ffunction-sections -fdata-sections -fno-guess-branch-probability \
+-funsigned-char \
+-falign-functions=1 -falign-jumps=1 -falign-labels=1 -falign-loops=1 \
+-march=i386 -mpreferred-stack-boundary=2 \
+-Wl,-Map -Wl,link.map -Wl,--warn-common -Wl,--sort-common -Wl,--gc-sections \
+httpd_indexcgi.c -o index.cgi
+
+Size (approximate):
+ text    data     bss     dec     hex filename
+22642     160    3052   25854    64fe index.cgi
+*/
+
+/* TODO: get rid of printf's: printf code is more than 50%
+ * of the entire executable when built against static uclibc */
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <errno.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <dirent.h>
+#include <time.h>
+
+/* Appearance of the table is controlled by style sheet *ONLY*,
+ * formatting code uses <TAG class=CLASS> to apply style
+ * to elements. Edit stylesheet to your liking and recompile. */
+
+static const char str_header[] =
+"" /* Additional headers (currently none) */
+"\r\n" /* Mandatory empty line after headers */
+"<html><head><title>Index of %s</title>"                "\n"
+"<style>"                                               "\n"
+"table {"                                               "\n"
+"  width: 100%%;"                                       "\n"
+"  background-color: #fff5ee;"                          "\n"
+"  border-width: 1px;" /* 1px 1px 1px 1px; */           "\n"
+"  border-spacing: 2px;"                                "\n"
+"  border-style: solid;" /* solid solid solid solid; */ "\n"
+"  border-color: black;" /* black black black black; */ "\n"
+"  border-collapse: collapse;"                          "\n"
+"}"                                                     "\n"
+"th {"                                                  "\n"
+"  border-width: 1px;" /* 1px 1px 1px 1px; */           "\n"
+"  padding: 1px;" /* 1px 1px 1px 1px; */                "\n"
+"  border-style: solid;" /* solid solid solid solid; */ "\n"
+"  border-color: black;" /* black black black black; */ "\n"
+"}"                                                     "\n"
+"td {"                                                  "\n"
+            /* top right bottom left */
+"  border-width: 0px 1px 0px 1px;"                      "\n"
+"  padding: 1px;" /* 1px 1px 1px 1px; */                "\n"
+"  border-style: solid;" /* solid solid solid solid; */ "\n"
+"  border-color: black;" /* black black black black; */ "\n"
+"}"                                                     "\n"
+"tr.hdr { background-color:#eee5de; }"                  "\n"
+"tr.o { background-color:#ffffff; }"                    "\n"
+/* tr.e { ... } - for even rows (currently none) */
+"tr.foot { background-color:#eee5de; }"                 "\n"
+"th.cnt { text-align:left; }"                           "\n"
+"th.sz { text-align:right; }"                           "\n"
+"th.dt { text-align:right; }"                           "\n"
+"td.sz { text-align:right; }"                           "\n"
+"td.dt { text-align:right; }"                           "\n"
+"col.nm { width: 98%%; }"                               "\n"
+"col.sz { width: 1%%; }"                                "\n"
+"col.dt { width: 1%%; }"                                "\n"
+"</style>"                                              "\n"
+"</head>"                                               "\n"
+"<body>"                                                "\n"
+"<h1>Index of %s</h1>"                                  "\n"
+""                                                      "\n"
+"<table>"                                               "\n"
+"<col class=nm><col class=sz><col class=dt>"            "\n"
+"<tr class=hdr><th class=cnt>Name<th class=sz>Size<th class=dt>Last modified" "\n"
+;
+
+static const char str_footer[] =
+"<tr class=foot><th class=cnt>Files: %u, directories: %u<th class=sz>%llu<th class=dt>&nbsp;" "\n"
+/* "</table></body></html>" - why bother? */
+;
+
+static int bad_url_char(unsigned c)
+{
+       return (c - '0') > 9 /* not a digit */
+           && ((c|0x20) - 'a') > 26 /* not A-Z or a-z */
+           && !strchr("._-+@", c);
+}
+
+static char *url_encode(const char *name)
+{
+       int i;
+       int size = 0;
+       int len = strlen(name);
+       char *p, *result;
+
+       i = 0;
+       while (name[i]) {
+               if (bad_url_char((unsigned)name[i]))
+                       size++;
+               i++;
+       }
+
+       /* No %xx needed! */
+       if (!size)
+               return (char*)name;
+
+       /* Each %xx requires 2 additional chars */
+       size = size * 2 + len + 1;
+       p = result = malloc(size);
+
+       i = 0;
+       while (name[i]) {
+               *p = name[i];
+               if (bad_url_char((unsigned)name[i])) {
+                       *p++ = '%';
+                       *p++ = "0123456789ABCDEF"[(uint8_t)(name[i]) >> 4];
+                       *p = "0123456789ABCDEF"[(uint8_t)(name[i]) & 0xf];
+               }
+               p++;
+               i++;
+       }
+       *p = 0;
+       return result;
+}
+
+static char *html_encode(const char *name)
+{
+       int i;
+       int size = 0;
+       int len = strlen(name);
+       char *p, *result;
+
+       i = 0;
+       while (name[i]) {
+               if (name[i] == '<'
+                || name[i] == '>'
+                || name[i] == '&'
+               ) {
+                       size++;
+               }
+               i++;
+       }
+
+       /* No &lt; etc needed! */
+       if (!size)
+               return (char*)name;
+
+       /* &amp; requires 4 additional chars */
+       size = size * 4 + len + 1;
+       p = result = malloc(size);
+
+       i = 0;
+       while (name[i]) {
+               char c;
+               *p = c = name[i++];
+               if (c == '<')
+                       strcpy(p, "&lt;");
+               else if (c == '>')
+                       strcpy(p, "&gt;");
+               else if (c == '&')
+                       strcpy(++p, "amp;");
+               else {
+                       p++;
+                       continue;
+               }
+               p += 4;
+       }
+       *p = 0;
+       return result;
+}
+
+typedef struct dir_list_t {
+       char  *dl_name;
+       mode_t dl_mode;
+       off_t  dl_size;
+       time_t dl_mtime;
+} dir_list_t;
+
+static int compare_dl(dir_list_t *a, dir_list_t *b)
+{
+       if (strcmp(a->dl_name, "..") == 0) {
+               /* ".." is 'less than' any other dir entry */
+               return -1;
+       }
+       if (strcmp(b->dl_name, "..") == 0) {
+               return 1;
+       }
+       if (S_ISDIR(a->dl_mode) != S_ISDIR(b->dl_mode)) {
+               /* 1 if b is a dir (and thus a is 'after' b, a > b),
+                * else -1 (a < b)*/
+               return (S_ISDIR(b->dl_mode) != 0) ? 1 : -1;
+       }
+       return strcmp(a->dl_name, b->dl_name);
+}
+
+int main(void)
+{
+       dir_list_t *dir_list;
+       dir_list_t *cdir;
+       unsigned dir_list_count;
+       unsigned count_dirs;
+       unsigned count_files;
+       unsigned long long size_total;
+       int odd;
+       DIR *dirp;
+       char *QUERY_STRING;
+
+       QUERY_STRING = getenv("QUERY_STRING");
+       if (!QUERY_STRING
+        || QUERY_STRING[0] != '/'
+        || strstr(QUERY_STRING, "/../")
+        || strcmp(strrchr(QUERY_STRING, '/'), "/..") == 0
+       ) {
+               return 1;
+       }
+
+       if (chdir("..")
+        || (QUERY_STRING[1] && chdir(QUERY_STRING + 1))
+       ) {
+               return 1;
+       }
+
+       dirp = opendir(".");
+       if (!dirp)
+               return 1;
+
+       dir_list = NULL;
+       dir_list_count = 0;
+       while (1) {
+               struct dirent *dp;
+               struct stat sb;
+
+               dp = readdir(dirp);
+               if (!dp)
+                       break;
+               if (dp->d_name[0] == '.' && !dp->d_name[1])
+                       continue;
+               if (stat(dp->d_name, &sb) != 0)
+                       continue;
+               dir_list = realloc(dir_list, (dir_list_count + 1) * sizeof(dir_list[0]));
+               dir_list[dir_list_count].dl_name = strdup(dp->d_name);
+               dir_list[dir_list_count].dl_mode = sb.st_mode;
+               dir_list[dir_list_count].dl_size = sb.st_size;
+               dir_list[dir_list_count].dl_mtime = sb.st_mtime;
+               dir_list_count++;
+       }
+
+       qsort(dir_list, dir_list_count, sizeof(dir_list[0]), (void*)compare_dl);
+
+       /* Guard against directories wit &, > etc */
+       QUERY_STRING = html_encode(QUERY_STRING);
+       printf(str_header, QUERY_STRING, QUERY_STRING);
+
+       odd = 0;
+       count_dirs = 0;
+       count_files = 0;
+       size_total = 0;
+
+       cdir = dir_list;
+       while (dir_list_count--) {
+               char size_str[sizeof(long long) * 3];
+               const char *slash_if_dir;
+               struct tm *tm;
+               char *href;
+               char *filename;
+               char datetime_str[sizeof("2000-02-02&nbsp;02:02:02")];
+
+               slash_if_dir = "/";
+               if (S_ISDIR(cdir->dl_mode)) {
+                       count_dirs++;
+                       size_str[0] = '\0';
+               } else if (S_ISREG(cdir->dl_mode)) {
+                       count_files++;
+                       size_total += cdir->dl_size;
+                       slash_if_dir++; /* points to "" now */
+                       sprintf(size_str, "%llu", (unsigned long long)(cdir->dl_size));
+               } else
+                       goto next;
+               href = url_encode(cdir->dl_name); /* %20 etc */
+               filename = html_encode(cdir->dl_name); /* &lt; etc */
+               tm = gmtime(&cdir->dl_mtime);
+               sprintf(datetime_str, "%04u-%02u-%02u&nbsp;%02u:%02u:%02u",
+                       (unsigned)(1900 + tm->tm_year),
+                       (unsigned)(tm->tm_mon + 1),
+                       (unsigned)(tm->tm_mday),
+                       (unsigned)(tm->tm_hour),
+                       (unsigned)(tm->tm_min),
+                       (unsigned)(tm->tm_sec)
+               );
+               printf("<tr class=%c><td class=nm><a href='%s%s'>%s%s</a><td class=sz>%s<td class=dt>%s\n",
+                       odd ? 'o' : 'e',
+                       href, slash_if_dir,
+                       filename, slash_if_dir,
+                       size_str,
+                       datetime_str
+               );
+               if (cdir->dl_name != href)
+                       free(href);
+               if (cdir->dl_name != filename)
+                       free(filename);
+               odd = 1 - odd;
+ next:
+               cdir++;
+       }
+
+       /* count_dirs - 1: we don't want to count ".." */
+       printf(str_footer, count_files, count_dirs - 1, size_total);
+       return 0;
+}