Add argument parsing

Added (IMHO) relatively simple argument parsing function (as well as helper
function for printing help information).

4 files changed, 694 insertions, 0 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 82e6123..5e0a380 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -67,6 +67,7 @@ endif(TESTING)
 # Set library code
 set(BH_SOURCE
     src/Algo.c
+    src/Args.c
     src/Hashmap.c
     src/IO.c
     src/Math.c
@@ -76,6 +77,7 @@ set(BH_SOURCE
 set(BH_HEADER
     include/BH/Common.h
     include/BH/Algo.h
+    include/BH/Args.h
     include/BH/Hashmap.h
     include/BH/IO.h
     include/BH/Math.h
diff --git a/include/BH/Args.h b/include/BH/Args.h
new file mode 100644
index 0000000..9932060
--- /dev/null
+++ b/include/BH/Args.h
@@ -0,0 +1,70 @@
+#ifndef BH_ARGS_H
+#define BH_ARGS_H
+
+
+#include "Common.h"
+
+
+#define BH_ARGS_ARGUMENT   0
+#define BH_ARGS_UNKNOWN   -1
+
+#define BH_ARGS_VALUE      1
+#define BH_ARGS_OPTIONAL   2
+
+
+typedef struct BH_ArgsOption
+{
+    int key;
+    const char *name;
+    int flags;
+    const char *description;
+} BH_ArgsOption;
+
+
+typedef int (*BH_ArgsCallback)(int key, char *arg, void *data);
+
+
+/**
+ * Parses command line \a options (given by \a argc and \a argv) and calls the
+ * specified \a callback with \a data.
+ *
+ * There are few internal rules:
+ * - If the "--" argument is specified, option parsing stops, and the rest of
+ *   the arguments are processed as is.
+ * - If the "-" argument is specified, it is processed as is.
+ * - For long options, if a value is required, the string after the equal sign
+ *   or the next argument will be used as the value ("--define=Value" or
+ *   "--define Value").
+ * - For short options, if a value is required, the string after the option
+ *   letter or the next argument will be used as the value ("-DValue" or
+ *   "-D Value").
+ *
+ * Options array should have last element filled with zeros.
+ *
+ * \param argc      Arguments count
+ * \param argv      Arguments array
+ * \param options   Options array pointer
+ * \param callback  Callback function
+ * \param data      Callback data
+ *
+ * \return On success, returns zero value.
+ * \return On failure, returns error code.
+ */
+int BH_ArgsParse(int argc,
+                 char **argv,
+                 BH_ArgsOption *options,
+                 BH_ArgsCallback callback,
+                 void *data);
+
+
+/**
+ * Prints documentation for \a options (with specified \a padding).
+ *
+ * \param options  Options array pointer
+ * \param padding  Padding
+ */
+void BH_ArgsHelp(BH_ArgsOption *options,
+                 int padding);
+
+
+#endif /* BH_ARGS_H */
\ No newline at end of file
diff --git a/src/Args.c b/src/Args.c
new file mode 100644
index 0000000..e780fed
--- /dev/null
+++ b/src/Args.c
@@ -0,0 +1,219 @@
+#include <BH/Args.h>
+#include <stdio.h>
+#include <ctype.h>
+#include <string.h>
+#include <memory.h>
+
+
+static int BH_ArgsExtractArg(int argc,
+                             char **argv,
+                             BH_ArgsOption *option,
+                             BH_ArgsCallback callback,
+                             void *data,
+                             int *i,
+                             int isLong,
+                             char *next)
+{
+    if (isLong && *next == '=')
+        return callback(option->key, next + 1, data);
+
+    if (!isLong && *next)
+        return callback(option->key, next, data);
+
+    if ((*i) + 1 < argc && !(argv[(*i) + 1][0] == '-' && argv[(*i) + 1][1]))
+        return callback(option->key, argv[++(*i)], data);
+
+    if (option->flags & BH_ARGS_OPTIONAL)
+        return callback(option->key, NULL, data);
+
+    return BH_ERROR;
+}
+
+
+static int BH_ArgsParseShort(int argc,
+                             char **argv,
+                             BH_ArgsOption *options,
+                             BH_ArgsCallback callback,
+                             void *data,
+                             int *i)
+{
+    char *symbol;
+    BH_ArgsOption *option;
+
+    for (symbol = argv[*i] + 1; *symbol; symbol++)
+    {
+        /* Find and process option */
+        for (option = options; option->key || option->name; option++)
+        {
+            if (option->key > 128 || *symbol != option->key)
+                continue;
+
+            if (option->flags & BH_ARGS_VALUE)
+                return BH_ArgsExtractArg(argc, argv, option, callback, data, i, 0, symbol + 1);
+
+            if (callback(option->key, NULL, data))
+                return BH_ERROR;
+
+            break;
+        }
+
+        /* Unknown option */
+        if (!option->key && !option->name)
+        {
+            callback(BH_ARGS_UNKNOWN, argv[*i], data);
+            return BH_ERROR;
+        }
+    }
+
+    return BH_OK;
+}
+
+
+static int BH_ArgsParseLong(int argc,
+                            char **argv,
+                            BH_ArgsOption *options,
+                            BH_ArgsCallback callback,
+                            void *data,
+                            int *i)
+{
+    char *start, *end;
+    BH_ArgsOption *option;
+
+    /* Extract option name from argv */
+    start = argv[*i] + 2;
+    end = argv[*i] + 2;
+    while (isgraph(*end) && *end != '=')
+        end++;
+
+    if (end == start)
+        return BH_ERROR;
+
+    /* Find and process option */
+    for (option = options; option->key || option->name; option++)
+    {
+        if (!option->name || strlen(option->name) != (size_t)(end - start))
+            continue;
+
+        if (memcmp(option->name, start, end - start))
+            continue;
+
+        if (option->flags & BH_ARGS_VALUE)
+            return BH_ArgsExtractArg(argc, argv, option, callback, data, i, 1, end);
+
+        return callback(option->key, NULL, data);
+    }
+
+    /* Unknown long option */
+    callback(BH_ARGS_UNKNOWN, argv[*i], data);
+    return BH_ERROR;
+}
+
+
+int BH_ArgsParse(int argc,
+                 char **argv,
+                 BH_ArgsOption *options,
+                 BH_ArgsCallback callback,
+                 void *data)
+{
+    int i, j, ignoreRest, size;
+    char *arg;
+
+    ignoreRest = 0; i = 1; size = argc;
+    while (i < argc)
+    {
+        arg = argv[i];
+
+        if (arg[0] == '-' && arg[1] && !ignoreRest)
+        {
+            /* Parse ingore, short or long option */
+            if (arg[1] == '-')
+            {
+                if (arg[2] && BH_ArgsParseLong(argc, argv, options, callback, data, &i))
+                    return BH_ERROR;
+                else if (!arg[2])
+                    ignoreRest = 1;
+
+            }
+            else if (arg[1] && BH_ArgsParseShort(argc, argv, options, callback, data, &i))
+                return BH_ERROR;
+
+            i++;
+        }
+        else
+        {
+            /* Shift arguments */
+            for (j = i; j < size - 1; j++)
+                argv[j] = argv[j + 1];
+            argv[size - 1] = arg;
+
+            if (callback(BH_ARGS_ARGUMENT, arg, data))
+                return BH_ERROR;
+
+            argc--;
+        }
+    }
+    return BH_OK;
+}
+
+
+void BH_ArgsHelp(BH_ArgsOption *options,
+                 int padding)
+{
+    int width = 0;
+    BH_ArgsOption *option;
+
+    if (padding == 0)
+    padding = 30;
+
+    /* Calculate option column width */
+    for (option = options; option->key || option->name; option++)
+    {
+        const char *symbol;
+        width = 5;
+
+        /* Print option column with padding */
+        if (isgraph(option->key))
+        {
+            fputs(" -", stdout);
+            fputc(option->key, stdout);
+        }
+        else
+            fputs("   ", stdout);
+
+        if (isgraph(option->key) && option->name)
+            fputs(", ", stdout);
+        else
+            fputs("  ", stdout);
+
+        if (option->name)
+        {
+            width += 3 + strlen(option->name);
+            fputs("--", stdout);
+            fputs(option->name, stdout);
+            fputc(' ', stdout);
+        }
+
+        if (option->flags & BH_ARGS_VALUE)
+        {
+            width += 8;
+            if (option->flags & BH_ARGS_OPTIONAL)
+            {
+                width += 2;
+                fputs("[<value>] ", stdout);
+            }
+            else
+                fputs("<value> ", stdout);
+        }
+        for (; width++ < padding; fputc(' ', stdout));
+
+        /* Print description */
+        for (symbol = option->description; symbol && *symbol; symbol++)
+        {
+            fputc(*symbol, stdout);
+
+            if (*symbol == '\n')
+                for (width = 0; width++ < padding; fputc(' ', stdout));
+        }
+        fputc('\n', stdout);
+    }
+}
diff --git a/test/src/TestArgs.c b/test/src/TestArgs.c
new file mode 100644
index 0000000..4c2f6ae
--- /dev/null
+++ b/test/src/TestArgs.c
@@ -0,0 +1,403 @@
+#include <BH/Args.h>
+#include <BH/Unit.h>
+#include <string.h>
+
+
+typedef struct DBG_State
+{
+    int key[128];
+    char *args[128];
+    size_t size;
+} DBG_State;
+
+
+static int DBG_Process(int key,
+                       char *arg,
+                       void *data)
+{
+    DBG_State *state = (DBG_State *)data;
+
+    state->key[state->size] = key;
+    state->args[state->size] = arg;
+    state->size++;
+
+    return BH_OK;
+}
+
+
+static int DBG_AlwaysPass(int key,
+                          char *arg,
+                          void *data)
+{
+    BH_UNUSED(key);
+    BH_UNUSED(arg);
+    BH_UNUSED(data);
+
+    return BH_OK;
+}
+
+
+static int DBG_AlwaysFail(int key,
+                          char *arg,
+                          void *data)
+{
+    BH_UNUSED(key);
+    BH_UNUSED(arg);
+    BH_UNUSED(data);
+
+    return BH_ERROR;
+}
+
+
+BH_UNIT_TEST(Empty)
+{
+    int argc = 1;
+
+    char *argv[] = {
+        "./app"
+    };
+
+    BH_ArgsOption options[] = {
+        {0, NULL, 0, NULL}
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_AlwaysFail, NULL) == BH_OK);
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_AlwaysPass, NULL) == BH_OK);
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) == BH_OK);
+    BH_VERIFY(state.size == 0);
+    BH_VERIFY(strcmp(argv[0], "./app") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(Args)
+{
+    int argc = 4;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "b",
+        "c"
+    };
+
+    BH_ArgsOption options[] = {
+        {0, NULL, 0, NULL}
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) == BH_OK);
+    BH_VERIFY(state.size == 3);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == 0 && strcmp(state.args[1], "b") == 0);
+    BH_VERIFY(state.key[2] == 0 && strcmp(state.args[2], "c") == 0);
+
+    BH_VERIFY(strcmp(argv[0], "./app") == 0);
+    BH_VERIFY(strcmp(argv[1], "a") == 0);
+    BH_VERIFY(strcmp(argv[2], "b") == 0);
+    BH_VERIFY(strcmp(argv[3], "c") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(Dash)
+{
+    int argc = 5;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "--",
+        "-a",
+        "--b"
+    };
+
+    BH_ArgsOption options[] = {
+        {0, NULL, 0, NULL}
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) == BH_OK);
+
+    BH_VERIFY(state.size == 3);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == 0 && strcmp(state.args[1], "-a") == 0);
+    BH_VERIFY(state.key[2] == 0 && strcmp(state.args[2], "--b") == 0);
+
+    BH_VERIFY(strcmp(argv[0], "./app") == 0);
+    BH_VERIFY(strcmp(argv[1], "--") == 0);
+    BH_VERIFY(strcmp(argv[2], "a") == 0);
+    BH_VERIFY(strcmp(argv[3], "-a") == 0);
+    BH_VERIFY(strcmp(argv[4], "--b") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(ShortOption)
+{
+    int argc = 12;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "-2",
+        "b",
+        "c",
+        "-2d",
+        "e",
+        "-2",
+        "-",
+        "f",
+        "-113",
+        "-1121",
+    };
+
+    BH_ArgsOption options[] = {
+        {'1', NULL, 0, NULL},
+        {'2', NULL, BH_ARGS_VALUE, NULL},
+        {'3', NULL, BH_ARGS_VALUE | BH_ARGS_OPTIONAL, NULL},
+        {0, NULL, 0, NULL},
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) == BH_OK);
+    BH_VERIFY(state.size == 13);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == '2' && strcmp(state.args[1], "b") == 0);
+    BH_VERIFY(state.key[2] == 0 && strcmp(state.args[2], "c") == 0);
+    BH_VERIFY(state.key[3] == '2' && strcmp(state.args[3], "d") == 0);
+    BH_VERIFY(state.key[4] == 0 && strcmp(state.args[4], "e") == 0);
+    BH_VERIFY(state.key[5] == '2' && strcmp(state.args[5], "-") == 0);
+    BH_VERIFY(state.key[6] == 0 && strcmp(state.args[6], "f") == 0);
+    BH_VERIFY(state.key[7] == '1' && state.args[7] == NULL);
+    BH_VERIFY(state.key[8] == '1' && state.args[8] == NULL);
+    BH_VERIFY(state.key[9] == '3' && state.args[9] == NULL);
+    BH_VERIFY(state.key[10] == '1' && state.args[10] == NULL);
+    BH_VERIFY(state.key[11] == '1' && state.args[11] == NULL);
+    BH_VERIFY(state.key[12] == '2' && strcmp(state.args[12], "1") == 0);
+
+    BH_VERIFY(strcmp(argv[0], "./app") == 0);
+    BH_VERIFY(strcmp(argv[1], "-2") == 0);
+    BH_VERIFY(strcmp(argv[2], "b") == 0);
+    BH_VERIFY(strcmp(argv[3], "-2d") == 0);
+    BH_VERIFY(strcmp(argv[4], "-2") == 0);
+    BH_VERIFY(strcmp(argv[5], "-") == 0);
+    BH_VERIFY(strcmp(argv[6], "-113") == 0);
+    BH_VERIFY(strcmp(argv[7], "-1121") == 0);
+    BH_VERIFY(strcmp(argv[8], "a") == 0);
+    BH_VERIFY(strcmp(argv[9], "c") == 0);
+    BH_VERIFY(strcmp(argv[10], "e") == 0);
+    BH_VERIFY(strcmp(argv[11], "f") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(LongOptions)
+{
+    int argc = 12;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "--def",
+        "b",
+        "c",
+        "--def=d",
+        "e",
+        "--def",
+        "-",
+        "f",
+        "--ghj",
+        "--abc",
+    };
+
+    BH_ArgsOption options[] = {
+        {1000, "abc", 0, NULL},
+        {1001, "def", BH_ARGS_VALUE, NULL},
+        {1002, "ghj", BH_ARGS_VALUE | BH_ARGS_OPTIONAL, NULL},
+        {0, NULL, 0, NULL},
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) == BH_OK);
+    BH_VERIFY(state.size == 9);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == 1001 && strcmp(state.args[1], "b") == 0);
+    BH_VERIFY(state.key[2] == 0 && strcmp(state.args[2], "c") == 0);
+    BH_VERIFY(state.key[3] == 1001 && strcmp(state.args[3], "d") == 0);
+    BH_VERIFY(state.key[4] == 0 && strcmp(state.args[4], "e") == 0);
+    BH_VERIFY(state.key[5] == 1001 && strcmp(state.args[5], "-") == 0);
+    BH_VERIFY(state.key[6] == 0 && strcmp(state.args[6], "f") == 0);
+    BH_VERIFY(state.key[7] == 1002 && state.args[7] == NULL);
+    BH_VERIFY(state.key[8] == 1000 && state.args[8] == NULL);
+
+    BH_VERIFY(strcmp(argv[0], "./app") == 0);
+    BH_VERIFY(strcmp(argv[1], "--def") == 0);
+    BH_VERIFY(strcmp(argv[2], "b") == 0);
+    BH_VERIFY(strcmp(argv[3], "--def=d") == 0);
+    BH_VERIFY(strcmp(argv[4], "--def") == 0);
+    BH_VERIFY(strcmp(argv[5], "-") == 0);
+    BH_VERIFY(strcmp(argv[6], "--ghj") == 0);
+    BH_VERIFY(strcmp(argv[7], "--abc") == 0);
+    BH_VERIFY(strcmp(argv[8], "a") == 0);
+    BH_VERIFY(strcmp(argv[9], "c") == 0);
+    BH_VERIFY(strcmp(argv[10], "e") == 0);
+    BH_VERIFY(strcmp(argv[11], "f") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(Complex)
+{
+    int argc = 15;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "-i",
+        "--input",
+        "b",
+        "-i-",
+        "-i",
+        "--input",
+        "-",
+        "--input",
+        "--",
+        "-a",
+        "-i",
+        "--input",
+        "--invalid",
+    };
+
+    BH_ArgsOption options[] = {
+        {'a', "abc", 0, NULL},
+        {'i', "input", BH_ARGS_VALUE | BH_ARGS_OPTIONAL, NULL},
+        {0, NULL, 0, NULL},
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) == BH_OK);
+
+    BH_VERIFY(state.size == 11);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == 'i' && state.args[1] == NULL);
+    BH_VERIFY(state.key[2] == 'i' && strcmp(state.args[2], "b") == 0);
+    BH_VERIFY(state.key[3] == 'i' && strcmp(state.args[3], "-") == 0);
+    BH_VERIFY(state.key[4] == 'i' && state.args[4] == NULL);
+    BH_VERIFY(state.key[5] == 'i' && strcmp(state.args[5], "-") == 0);
+    BH_VERIFY(state.key[6] == 'i' && state.args[6] == NULL);
+    BH_VERIFY(state.key[7] == 0 && strcmp(state.args[7], "-a") == 0);
+    BH_VERIFY(state.key[8] == 0 && strcmp(state.args[8], "-i") == 0);
+    BH_VERIFY(state.key[9] == 0 && strcmp(state.args[9], "--input") == 0);
+    BH_VERIFY(state.key[10] == 0 && strcmp(state.args[10], "--invalid") == 0);
+
+    BH_VERIFY(strcmp(argv[0], "./app") == 0);
+    BH_VERIFY(strcmp(argv[1], "-i") == 0);
+    BH_VERIFY(strcmp(argv[2], "--input") == 0);
+    BH_VERIFY(strcmp(argv[3], "b") == 0);
+    BH_VERIFY(strcmp(argv[4], "-i-") == 0);
+    BH_VERIFY(strcmp(argv[5], "-i") == 0);
+    BH_VERIFY(strcmp(argv[6], "--input") == 0);
+    BH_VERIFY(strcmp(argv[7], "-") == 0);
+    BH_VERIFY(strcmp(argv[8], "--input") == 0);
+    BH_VERIFY(strcmp(argv[9], "--") == 0);
+    BH_VERIFY(strcmp(argv[10], "a") == 0);
+    BH_VERIFY(strcmp(argv[11], "-a") == 0);
+    BH_VERIFY(strcmp(argv[12], "-i") == 0);
+    BH_VERIFY(strcmp(argv[13], "--input") == 0);
+    BH_VERIFY(strcmp(argv[14], "--invalid") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(InvalidShort)
+{
+    int argc = 4;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "-b",
+        "-c"
+    };
+
+    BH_ArgsOption options[] = {
+        {0, NULL, 0, NULL}
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) != BH_OK);
+
+    BH_VERIFY(state.size == 2);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == BH_ARGS_UNKNOWN && strcmp(state.args[1], "-b") == 0);
+
+    return 0;
+}
+
+
+BH_UNIT_TEST(InvalidLong)
+{
+    int argc = 4;
+
+    char *argv[] = {
+        "./app",
+        "a",
+        "--b",
+        "--c"
+    };
+
+    BH_ArgsOption options[] = {
+        {0, NULL, 0, NULL}
+    };
+
+    DBG_State state;
+    memset(&state, 0, sizeof(state));
+
+    BH_VERIFY(BH_ArgsParse(argc, argv, options, DBG_Process, &state) != BH_OK);
+
+    BH_VERIFY(state.size == 2);
+    BH_VERIFY(state.key[0] == 0 && strcmp(state.args[0], "a") == 0);
+    BH_VERIFY(state.key[1] == BH_ARGS_UNKNOWN && strcmp(state.args[1], "--b") == 0);
+
+    return 0;
+}
+
+
+int main(int argc, char **argv)
+{
+    (void)argc;
+    (void)argv;
+
+    BH_UNIT_ADD(Empty);
+    BH_UNIT_ADD(Args);
+    BH_UNIT_ADD(Dash);
+    BH_UNIT_ADD(ShortOption);
+    BH_UNIT_ADD(LongOptions);
+    BH_UNIT_ADD(Complex);
+    BH_UNIT_ADD(InvalidShort);
+    BH_UNIT_ADD(InvalidLong);
+
+    return BH_UnitRun();
+}