git-receiver: splitted/reimplemented and tested shell command parser

7 files changed, 448 insertions, 56 deletions

diff --git a/.gitignore b/.gitignore
index af2b5d8..98aee66 100644
--- a/.gitignore
+++ b/.gitignore
@@ -58,6 +58,7 @@ blogc*.html
 /tests/blogc/check_source_parser
 /tests/blogc/check_template_parser
 /tests/blogc-git-receiver/check_pre_receive_parser
+/tests/blogc-git-receiver/check_shell_command_parser
 /tests/blogc-runserver/check_httpd_utils
 /tests/blogc-runserver/check_mime
 /tests/common/check_config_parser
diff --git a/Makefile.am b/Makefile.am
index fff49b9..31c1dca 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -43,6 +43,7 @@ noinst_HEADERS = \
 	src/blogc-git-receiver/pre-receive.h \
 	src/blogc-git-receiver/pre-receive-parser.h \
 	src/blogc-git-receiver/shell.h \
+	src/blogc-git-receiver/shell-command-parser.h \
 	src/blogc-runserver/httpd.h \
 	src/blogc-runserver/httpd-utils.h \
 	src/blogc-runserver/mime.h \
@@ -156,6 +157,7 @@ libblogc_git_receiver_la_SOURCES = \
 	src/blogc-git-receiver/pre-receive.c \
 	src/blogc-git-receiver/pre-receive-parser.c \
 	src/blogc-git-receiver/shell.c \
+	src/blogc-git-receiver/shell-command-parser.c \
 	$(NULL)
 
 libblogc_git_receiver_la_CFLAGS = \
@@ -571,6 +573,7 @@ endif
 if BUILD_GIT_RECEIVER
 check_PROGRAMS += \
 	tests/blogc-git-receiver/check_pre_receive_parser \
+	tests/blogc-git-receiver/check_shell_command_parser \
 	$(NULL)
 
 tests_blogc_git_receiver_check_pre_receive_parser_SOURCES = \
@@ -590,6 +593,24 @@ tests_blogc_git_receiver_check_pre_receive_parser_LDADD = \
 	libblogc_git_receiver.la \
 	libblogc_common.la \
 	$(NULL)
+
+tests_blogc_git_receiver_check_shell_command_parser_SOURCES = \
+	tests/blogc-git-receiver/check_shell_command_parser.c \
+	$(NULL)
+
+tests_blogc_git_receiver_check_shell_command_parser_CFLAGS = \
+	$(CMOCKA_CFLAGS) \
+	$(NULL)
+
+tests_blogc_git_receiver_check_shell_command_parser_LDFLAGS = \
+	-no-install \
+	$(NULL)
+
+tests_blogc_git_receiver_check_shell_command_parser_LDADD = \
+	$(CMOCKA_LIBS) \
+	libblogc_git_receiver.la \
+	libblogc_common.la \
+	$(NULL)
 endif
 
 endif
diff --git a/src/blogc-git-receiver/shell-command-parser.c b/src/blogc-git-receiver/shell-command-parser.c
new file mode 100644
index 0000000..cc8c537
--- /dev/null
+++ b/src/blogc-git-receiver/shell-command-parser.c
@@ -0,0 +1,125 @@
+/*
+ * blogc: A blog compiler.
+ * Copyright (C) 2015-2016 Rafael G. Martins <rafael@rafaelmartins.eng.br>
+ *
+ * This program can be distributed under the terms of the BSD License.
+ * See the file LICENSE.
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include "../common/utils.h"
+#include "shell-command-parser.h"
+
+typedef enum {
+    START_COMMAND = 1,
+    START_REPO,
+    START_REPO2,
+    REPO,
+    START_ESCAPED,
+} command_state_t;
+
+
+char*
+bgr_shell_command_parse(const char *command)
+{
+    command_state_t state = START_COMMAND;
+    size_t start = 0;
+    size_t command_len = strlen(command);
+
+    bc_string_t *rv = bc_string_new();
+
+    for (size_t current = 0; current < command_len; current++) {
+
+        char c = command[current];
+
+        switch (state) {
+            case START_COMMAND:
+                if (c == ' ') {
+                    if (((current == 16) &&
+                         (0 == strncmp("git-receive-pack", command, 16))) ||
+                        ((current == 15) &&
+                         (0 == strncmp("git-upload-pack", command, 15))) ||
+                        ((current == 18) &&
+                         (0 == strncmp("git-upload-archive", command, 18))))
+                    {
+                        state = START_REPO;
+                        break;
+                    }
+                    goto error;
+                }
+                break;
+
+            case START_REPO:
+                if (c == '\'') {  // never saw git using double-quotes
+                    state = START_REPO2;
+                    break;
+                }
+                if (c == '\\') {  // escaped ! or '
+                    state = START_ESCAPED;
+                    break;
+                }
+                goto error;
+
+            case START_REPO2:
+                if (c == '\'') {
+                    state = START_REPO;
+                    break;
+                }
+                start = current;
+                if (rv->len == 0 && c == '/') {  // no absolute urls
+                    start = current + 1;
+                }
+                state = REPO;
+                break;
+
+            case START_ESCAPED:
+                if (c == '!' || c == '\'') {
+                    bc_string_append_c(rv, c);
+                    state = START_REPO;
+                    break;
+                }
+                goto error;
+
+            case REPO:
+                if (c == '\'') {
+                    bc_string_append_len(rv, command + start, current - start);
+                    state = START_REPO;
+                    break;
+                }
+                break;
+        }
+    }
+
+    if (rv->len > 0)
+        return bc_string_free(rv, false);
+
+error:
+    bc_string_free(rv, true);
+    return NULL;
+}
+
+
+char*
+bgr_shell_quote(const char *command)
+{
+    // this does not really belongs here, but function is very small
+    bc_string_t *rv = bc_string_new();
+    bc_string_append_c(rv, '\'');
+    if (command != NULL) {
+        for (size_t i = 0; i < strlen(command); i++) {
+            switch (command[i]) {
+                case '!':
+                    bc_string_append(rv, "'\\!'");
+                    break;
+                case '\'':
+                    bc_string_append(rv, "'\\''");
+                    break;
+                default:
+                    bc_string_append_c(rv, command[i]);
+            }
+        }
+    }
+    bc_string_append_c(rv, '\'');
+    return bc_string_free(rv, false);
+}
diff --git a/src/blogc-git-receiver/shell-command-parser.h b/src/blogc-git-receiver/shell-command-parser.h
new file mode 100644
index 0000000..47054cb
--- /dev/null
+++ b/src/blogc-git-receiver/shell-command-parser.h
@@ -0,0 +1,15 @@
+/*
+ * blogc: A blog compiler.
+ * Copyright (C) 2015-2016 Rafael G. Martins <rafael@rafaelmartins.eng.br>
+ *
+ * This program can be distributed under the terms of the BSD License.
+ * See the file LICENSE.
+ */
+
+#ifndef _SHELL_COMMAND_PARSER_H
+#define _SHELL_COMMAND_PARSER_H
+
+char* bgr_shell_command_parse(const char *command);
+char* bgr_shell_quote(const char *command);
+
+#endif /* _SHELL_COMMAND_PARSER_H */
diff --git a/src/blogc-git-receiver/shell.c b/src/blogc-git-receiver/shell.c
index 290526b..581ec3d 100644
--- a/src/blogc-git-receiver/shell.c
+++ b/src/blogc-git-receiver/shell.c
@@ -16,12 +16,9 @@
 #include <sys/stat.h>
 #include <sys/types.h>
 #include "../common/utils.h"
+#include "shell-command-parser.h"
 #include "shell.h"
 
-#ifndef BUFFER_SIZE
-#define BUFFER_SIZE 4096
-#endif
-
 
 int
 bgr_shell(int argc, char *argv[])
@@ -29,22 +26,7 @@ bgr_shell(int argc, char *argv[])
     int rv = 0;
 
     char *repo = NULL;
-    char *command_orig = NULL;
-    char *command_name = NULL;
-    char command_new[BUFFER_SIZE];
-
-    bool exec_git = false;
-
-    // validate git command
-    size_t len = strlen(argv[2]);
-    if (!((len > 17 && (0 == strncmp(argv[2], "git-receive-pack ", 17))) ||
-          (len > 16 && (0 == strncmp(argv[2], "git-upload-pack ", 16))) ||
-          (len > 19 && (0 == strncmp(argv[2], "git-upload-archive ", 19)))))
-    {
-        fprintf(stderr, "error: unsupported git command: %s\n", argv[2]);
-        rv = 1;
-        goto cleanup;
-    }
+    char *quoted_repo = NULL;
 
     // get shell path
     char *self = getenv("SHELL");
@@ -62,23 +44,17 @@ bgr_shell(int argc, char *argv[])
         goto cleanup;
     }
 
-    // get git repository
-    command_orig = bc_strdup(argv[2]);
-    char *p, *r;
-    for (p = command_orig; *p != ' ' && *p != '\0'; p++);
-    if (*p == ' ')
-        p++;
-    if (*p == '\'' || *p == '"')
-        p++;
-    if (*p == '/')
-        p++;
-    for (r = p; *p != '\'' && *p != '"' && *p != '\0'; p++);
-    if (*p == '\'' || *p == '"')
-        *p = '\0';
-    if (*--p == '/')
-        *p = '\0';
-
-    repo = bc_strdup_printf("repos/%s", r);
+    // validate command and extract git repository
+    char *tmp_repo = bgr_shell_command_parse(argv[2]);
+    if (tmp_repo == NULL) {
+        fprintf(stderr, "error: invalid git-shell command: %s\n", argv[2]);
+        rv = 1;
+        goto cleanup;
+    }
+
+    repo = bc_strdup_printf("repos/%s", tmp_repo);
+    quoted_repo = bgr_shell_quote(repo);
+    free(tmp_repo);
 
     // check if repository is sane
     if (0 == strlen(repo)) {
@@ -99,7 +75,7 @@ bgr_shell(int argc, char *argv[])
 
     if (0 != access(repo, F_OK)) {
         char *git_init_cmd = bc_strdup_printf(
-            "git init --bare \"%s\" > /dev/null", repo);
+            "git init --bare %s > /dev/null", quoted_repo);
         if (0 != system(git_init_cmd)) {
             fprintf(stderr, "error: failed to create git repository: %s\n",
                 repo);
@@ -175,32 +151,41 @@ bgr_shell(int argc, char *argv[])
     }
 
 git_exec:
-    command_name = bc_strdup(argv[2]);
-    for (p = command_name; *p != ' ' && *p != '\0'; p++);
-    if (*p == ' ')
-        *p = '\0';
 
-    if (BUFFER_SIZE < (strlen(command_name) + strlen(repo) + 4)) {
-        fprintf(stderr, "error: git-shell command is too big\n");
-        rv = 1;
-        goto cleanup;
-    }
+    {
+        // static allocation instead of bc_strdup_printf to avoid leaks
+        char buffer[4096];
+        char *command = bc_strdup(argv[2]);
+        char *p;
+        for (p = command; *p != ' ' && *p != '\0'; p++);
+        if (*p == ' ')
+            *p = '\0';
+
+        if (sizeof(buffer) < (strlen(command) + strlen(quoted_repo) + 2)) {
+            fprintf(stderr, "error: git-shell command is too big\n");
+            rv = 1;
+            goto cleanup;
+        }
 
-    if (snprintf(command_new, BUFFER_SIZE, "%s '%s'", command_name, repo))
-        exec_git = true;
+        if (0 > snprintf(buffer, sizeof(buffer), "%s %s", command, quoted_repo)) {
+            fprintf(stderr, "error: failed to generate git-shell command\n");
+            rv = 1;
+            goto cleanup;
+        }
 
-cleanup:
-    free(repo);
-    free(command_orig);
-    free(command_name);
+        free(command);
+        free(repo);
+        free(quoted_repo);
 
-    if (exec_git) {
-        execlp("git-shell", "git-shell", "-c", command_new, NULL);
+        execlp("git-shell", "git-shell", "-c", buffer, NULL);
 
         // execlp only returns on error, then something bad happened
         fprintf(stderr, "error: failed to execute git-shell\n");
-        rv = 1;
+        return 1;  // avoid freeing repo again
     }
 
+cleanup:
+    free(repo);
+    free(quoted_repo);
     return rv;
 }
diff --git a/src/blogc-git-receiver/shell.h b/src/blogc-git-receiver/shell.h
index c05e4e1..ec77db1 100644
--- a/src/blogc-git-receiver/shell.h
+++ b/src/blogc-git-receiver/shell.h
@@ -9,6 +9,7 @@
 #ifndef _SHELL_H
 #define _SHELL_H
 
+char* bgr_shell_get_repo(const char *command);
 int bgr_shell(int argc, char *argv[]);
 
 #endif /* _SHELL_H */
diff --git a/tests/blogc-git-receiver/check_shell_command_parser.c b/tests/blogc-git-receiver/check_shell_command_parser.c
new file mode 100644
index 0000000..9d0e282
--- /dev/null
+++ b/tests/blogc-git-receiver/check_shell_command_parser.c
@@ -0,0 +1,244 @@
+/*
+ * blogc: A blog compiler.
+ * Copyright (C) 2015-2016 Rafael G. Martins <rafael@rafaelmartins.eng.br>
+ *
+ * This program can be distributed under the terms of the BSD License.
+ * See the file LICENSE.
+ */
+
+#include <stdarg.h>
+#include <stddef.h>
+#include <setjmp.h>
+#include <cmocka.h>
+#include <stdlib.h>
+#include "../../src/blogc-git-receiver/shell-command-parser.h"
+
+
+static void
+test_shell_command_parse(void **state)
+{
+    char *t;
+    assert_null(bgr_shell_command_parse(""));
+    assert_null(bgr_shell_command_parse("bola"));
+    assert_null(bgr_shell_command_parse("bola guda"));
+    assert_null(bgr_shell_command_parse("bola 'guda'"));
+    t = bgr_shell_command_parse("git-receive-pack 'bola.git'");
+    assert_string_equal(t, "bola.git");
+    free(t);
+    t = bgr_shell_command_parse("git-upload-pack 'bolaa.git'");
+    assert_string_equal(t, "bolaa.git");
+    free(t);
+    t = bgr_shell_command_parse("git-upload-archive 'bolab.git'");
+    assert_string_equal(t, "bolab.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bola1.git'");
+    assert_string_equal(t, "bola1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bola2.git/'");
+    assert_string_equal(t, "bola2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'bola3.git/'");
+    assert_string_equal(t, "bola3.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bola.git'");
+    assert_string_equal(t, "foo/bola.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bola1.git'");
+    assert_string_equal(t, "foo/bola1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bola2.git/'");
+    assert_string_equal(t, "foo/bola2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bola3.git/'");
+    assert_string_equal(t, "foo/bola3.git/");
+    free(t);
+
+    t = bgr_shell_command_parse("git-receive-pack ''\\''bola.git'");
+    assert_string_equal(t, "'bola.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''/bola1.git'");
+    assert_string_equal(t, "'/bola1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''/bola2.git/'");
+    assert_string_equal(t, "'/bola2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''bola3.git/'");
+    assert_string_equal(t, "'bola3.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''foo/bola.git'");
+    assert_string_equal(t, "'foo/bola.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''/foo/bola1.git'");
+    assert_string_equal(t, "'/foo/bola1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''/foo/bola2.git/'");
+    assert_string_equal(t, "'/foo/bola2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\''foo/bola3.git/'");
+    assert_string_equal(t, "'foo/bola3.git/");
+    free(t);
+
+    t = bgr_shell_command_parse("git-receive-pack 'bola.git'\\'''");
+    assert_string_equal(t, "bola.git'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bola1.git'\\'''");
+    assert_string_equal(t, "bola1.git'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bola2.git/'\\'''");
+    assert_string_equal(t, "bola2.git/'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'bola3.git/'\\'''");
+    assert_string_equal(t, "bola3.git/'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bola.git'\\'''");
+    assert_string_equal(t, "foo/bola.git'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bola1.git'\\'''");
+    assert_string_equal(t, "foo/bola1.git'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bola2.git/'\\'''");
+    assert_string_equal(t, "foo/bola2.git/'");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bola3.git/'\\'''");
+    assert_string_equal(t, "foo/bola3.git/'");
+    free(t);
+
+    t = bgr_shell_command_parse("git-receive-pack 'bo'\\''la.git'");
+    assert_string_equal(t, "bo'la.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bo'\\''la1.git'");
+    assert_string_equal(t, "bo'la1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bo'\\''la2.git/'");
+    assert_string_equal(t, "bo'la2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'bo'\\''la3.git/'");
+    assert_string_equal(t, "bo'la3.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bo'\\''la.git'");
+    assert_string_equal(t, "foo/bo'la.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bo'\\''la1.git'");
+    assert_string_equal(t, "foo/bo'la1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bo'\\''la2.git/'");
+    assert_string_equal(t, "foo/bo'la2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bo'\\''la3.git/'");
+    assert_string_equal(t, "foo/bo'la3.git/");
+    free(t);
+
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'bola.git'");
+    assert_string_equal(t, "!bola.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'/bola1.git'");
+    assert_string_equal(t, "!/bola1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'/bola2.git/'");
+    assert_string_equal(t, "!/bola2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'bola3.git/'");
+    assert_string_equal(t, "!bola3.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'foo/bola.git'");
+    assert_string_equal(t, "!foo/bola.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'/foo/bola1.git'");
+    assert_string_equal(t, "!/foo/bola1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'/foo/bola2.git/'");
+    assert_string_equal(t, "!/foo/bola2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack ''\\!'foo/bola3.git/'");
+    assert_string_equal(t, "!foo/bola3.git/");
+    free(t);
+
+    t = bgr_shell_command_parse("git-receive-pack 'bola.git'\\!''");
+    assert_string_equal(t, "bola.git!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bola1.git'\\!''");
+    assert_string_equal(t, "bola1.git!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bola2.git/'\\!''");
+    assert_string_equal(t, "bola2.git/!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'bola3.git/'\\!''");
+    assert_string_equal(t, "bola3.git/!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bola.git'\\!''");
+    assert_string_equal(t, "foo/bola.git!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bola1.git'\\!''");
+    assert_string_equal(t, "foo/bola1.git!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bola2.git/'\\!''");
+    assert_string_equal(t, "foo/bola2.git/!");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bola3.git/'\\!''");
+    assert_string_equal(t, "foo/bola3.git/!");
+    free(t);
+
+    t = bgr_shell_command_parse("git-receive-pack 'bo'\\!'la.git'");
+    assert_string_equal(t, "bo!la.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bo'\\!'la1.git'");
+    assert_string_equal(t, "bo!la1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/bo'\\!'la2.git/'");
+    assert_string_equal(t, "bo!la2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'bo'\\!'la3.git/'");
+    assert_string_equal(t, "bo!la3.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bo'\\!'la.git'");
+    assert_string_equal(t, "foo/bo!la.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bo'\\!'la1.git'");
+    assert_string_equal(t, "foo/bo!la1.git");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack '/foo/bo'\\!'la2.git/'");
+    assert_string_equal(t, "foo/bo!la2.git/");
+    free(t);
+    t = bgr_shell_command_parse("git-receive-pack 'foo/bo'\\!'la3.git/'");
+    assert_string_equal(t, "foo/bo!la3.git/");
+    free(t);
+}
+
+
+static void
+test_shell_quote(void **state)
+{
+    char *t;
+    t = bgr_shell_quote(NULL);
+    assert_string_equal(t, "''");
+    free(t);
+    t = bgr_shell_quote("!bola");
+    assert_string_equal(t, "''\\!'bola'");
+    free(t);
+    t = bgr_shell_quote("'bola");
+    assert_string_equal(t, "''\\''bola'");
+    free(t);
+    t = bgr_shell_quote("bo!bola");
+    assert_string_equal(t, "'bo'\\!'bola'");
+    free(t);
+    t = bgr_shell_quote("bo'bola");
+    assert_string_equal(t, "'bo'\\''bola'");
+    free(t);
+    t = bgr_shell_quote("bola!");
+    assert_string_equal(t, "'bola'\\!''");
+    free(t);
+    t = bgr_shell_quote("bola'");
+    assert_string_equal(t, "'bola'\\'''");
+    free(t);
+}
+
+
+int
+main(void)
+{
+    const UnitTest tests[] = {
+        unit_test(test_shell_command_parse),
+        unit_test(test_shell_quote),
+    };
+    return run_tests(tests);
+}