Reading xref section (with all subsections)

added
  - tests
  - fn add_xref_entry
  - fn free_xref_entry
  - fn xref_init
  - fn xref_free
  - fn read_xref_table
  - fn print_xref
  - fn skip_leading_whitespaces
  - fn parse_keyword
removed
  - modes (also sigil_process -> sigil_verify)

include/auxiliary.h

@@ -2,12 +2,33 @@
 #define PDF_SIGIL_AUXILIARY_H
 
 #include <stdio.h>
+#include <stdint.h>
 
 #ifndef CHAR_T
 #define CHAR_T
-    typedef char char_t;
+    typedef unsigned char char_t;
 #endif /* CHAR_T */
 
+#ifndef SIGIL_ERR_T
+#define SIGIL_ERR_T
+    typedef uint32_t sigil_err_t;
+#endif /* SIGIL_ERR_T */
+
+#ifndef KEYWORD_T
+#define KEYWORD_T
+    typedef uint32_t keyword_t;
+#endif /* KEYWORD_T */
+
+#ifndef FREE_INDICATOR_T
+#define FREE_INDICATOR_T
+    typedef uint32_t free_indicator_t;
+#endif /* FREE_INDICATOR_T */
+
+#define KEYWORD_xref     0
+#define KEYWORD_trailer  1
+
+#define IN_USE_ENTRY     0
+#define FREE_ENTRY       1
 
 #define COLOR_RED        "\x1b[31m"
 #define COLOR_GREEN      "\x1b[32m"
@@ -22,8 +43,11 @@ void sigil_zeroize(void *a, size_t bytes);
 int is_digit(const char_t c);
 int is_whitespace(const char_t c);
 
-int parse_number(FILE *in, size_t *number);
-int parse_free_indicator(FILE *in, char_t *result);
+sigil_err_t skip_leading_whitespaces(FILE *in);
+
+sigil_err_t parse_number(FILE *in, size_t *number);
+sigil_err_t parse_keyword(FILE *in, keyword_t *keyword);
+sigil_err_t parse_free_indicator(FILE *in, free_indicator_t *result);
 
 void print_module_name(const char *module_name, int verbosity);
 void print_module_result(int result, int verbosity);

include/error.h

@@ -5,9 +5,13 @@
 
 #ifndef CHAR_T
 #define CHAR_T
-    typedef char char_t;
+    typedef unsigned char char_t;
 #endif /* CHAR_T */
 
+#ifndef SIGIL_ERR_T
+#define SIGIL_ERR_T
+    typedef uint32_t sigil_err_t;
+#endif /* SIGIL_ERR_T */
 
 #define ERR_NO           0x0000 // [_|_|_|x] 0000 0000
 #define ERR_ALLOC        0x0001 // [_|_|_|x] 0000 0001
@@ -27,9 +31,7 @@
 #define ERR_15           0x4000 // [_|_|x|_] 0100 0000
 #define ERR_16           0x8000 // [_|_|x|_] 1000 0000
 
-typedef uint32_t sigil_err_t;
-
-const char_t *sigil_err_string(sigil_err_t err);
+const char *sigil_err_string(sigil_err_t err);
 
 int sigil_error_self_test(int verbosity);
 

include/header.h

 #ifndef PDF_SIGIL_HEADER_H
 #define PDF_SIGIL_HEADER_H
 
-#include "error.h"
 #include "sigil.h"
 
+#ifndef SIGIL_ERR_T
+#define SIGIL_ERR_T
+    typedef uint32_t sigil_err_t;
+#endif /* SIGIL_ERR_T */
+
 
 sigil_err_t process_header(sigil_t *sgl);
 

include/sigil.h

@@ -2,13 +2,17 @@
 #define PDF_SIGIL_SIGIL_H
 
 #include <stdio.h>
-#include "error.h"
 
 #ifndef CHAR_T
 #define CHAR_T
-    typedef char char_t;
+    typedef unsigned char char_t;
 #endif /* CHAR_T */
 
+#ifndef SIGIL_ERR_T
+#define SIGIL_ERR_T
+    typedef uint32_t sigil_err_t;
+#endif /* SIGIL_ERR_T */
+
 #ifndef XREF_T
 #define XREF_T
     typedef struct {
@@ -26,17 +30,9 @@
 #define XREF_TYPE_TABLE    1
 #define XREF_TYPE_STREAM   2
 
-#define MODE_UNSET     0
-#define MODE_VERIFY    1
-#define MODE_SIGN      2
-
-typedef uint32_t mode_t;
-struct xref_t;
-
 typedef struct {
     FILE   *file;
-    char_t *filepath;
-    mode_t  mode;
+    char   *filepath;
     short   pdf_x,             /* numbers from PDF header */
             pdf_y;             /*   %PDF-<pdf_x>.<pdf_y>  */
     short   xref_type;
@@ -48,13 +44,11 @@ typedef struct {
 
 sigil_err_t sigil_init(sigil_t **sgl);
 
-sigil_err_t sigil_setup_file(sigil_t *sgl, const char_t *filepath);
-
-sigil_err_t sigil_setup_mode(sigil_t *sgl, mode_t mode);
+sigil_err_t sigil_setup_file(sigil_t *sgl, const char *filepath);
 
-sigil_err_t sigil_process(sigil_t *sgl);
+sigil_err_t sigil_verify(sigil_t *sgl);
 
-// ... get functions TBD
+// ... get functions TODO
 
 void sigil_free(sigil_t *sgl);
 

include/trailer.h

 #ifndef PDF_SIGIL_TRAILER_H
 #define PDF_SIGIL_TRAILER_H
 
-#include "error.h"
+#ifndef SIGIL_ERR_T
+#define SIGIL_ERR_T
+    typedef uint32_t sigil_err_t;
+#endif /* SIGIL_ERR_T */
 
 
 sigil_err_t process_trailer(sigil_t *sgl);

include/xref.h

 #ifndef PDF_SIGIL_XREF_H
 #define PDF_SIGIL_XREF_H
 
-#include "error.h"
 #include "sigil.h"
 
+#ifndef SIGIL_ERR_T
+#define SIGIL_ERR_T
+    typedef uint32_t sigil_err_t;
+#endif /* SIGIL_ERR_T */
 
 #ifndef XREF_T
 #define XREF_T
-
-typedef struct {
-    size_t byte_offset;
-    size_t generation_num;
-} xref_entry_t;
-
-typedef struct {
-    xref_entry_t **entry;
-    size_t capacity;
-} xref_t;
-
+    typedef struct {
+        size_t byte_offset;
+        size_t generation_num;
+    } xref_entry_t;
+
+    typedef struct {
+        xref_entry_t **entry;
+        size_t capacity;
+    } xref_t;
 #endif /* XREF_T */
 
-xref_t *new_xref();
+xref_t *xref_init();
 
-void free_xref(xref_t *xref);
+void xref_free(xref_t *xref);
 
 sigil_err_t read_startxref(sigil_t *sgl);
 

lib/auxiliary.c

 #include <stdlib.h>
 #include <stdio.h>
+#include <string.h>
 #include "auxiliary.h"
+#include "error.h"
 
 void sigil_zeroize(void *a, size_t bytes)
 {
@@ -29,46 +31,113 @@ int is_whitespace(const char_t c)
             c == 0x20);  // space
 }
 
-int parse_number(FILE *in, size_t *number)
+sigil_err_t skip_leading_whitespaces(FILE *in)
 {
-    char c;
+    char_t c;
+
+    while ((c = fgetc(in)) != EOF && is_whitespace(c))
+        ;
+
+    if (c == EOF)
+        return (sigil_err_t)ERR_PDF_CONT;
+
+    if (ungetc(c, in) != c)
+        return (sigil_err_t)ERR_IO;
+
+    return (sigil_err_t)ERR_NO;
+}
+
+sigil_err_t parse_number(FILE *in, size_t *number)
+{
+    char_t c;
     int digits = 0;
+    sigil_err_t err;
 
     *number = 0;
 
-    // skip leading whitespaces
-    while ((c = fgetc(in)) != EOF && is_whitespace(c))
-        ;
+    err = skip_leading_whitespaces(in);
+    if (err != ERR_NO)
+        return err;
 
     // number
-    do {
+    while ((c = fgetc(in)) != EOF) {
         if (!is_digit(c)) {
             if (ungetc(c, in) != c)
-                return 1;
-            return digits == 0;
+                return (sigil_err_t)ERR_IO;
+            if (digits > 0) {
+                return (sigil_err_t)ERR_NO;
+            } else {
+                return (sigil_err_t)ERR_PDF_CONT;
+            }
         }
         *number = 10 * *number + c - '0';
         digits++;
-    } while ((c = fgetc(in)) != EOF);
+    }
 
-    return 1;
+    return (sigil_err_t)ERR_PDF_CONT;
 }
 
-int parse_free_indicator(FILE *in, char_t *result)
+sigil_err_t parse_keyword(FILE *in, keyword_t *keyword)
 {
+    sigil_err_t err;
+    int count = 0;
+    const int keyword_max = 10;
+    char tmp[keyword_max],
+         c;
+
+    sigil_zeroize(tmp, keyword_max * sizeof(*tmp));
+
+    err = skip_leading_whitespaces(in);
+    if (err != ERR_NO)
+        return err;
+
+    while ((c = fgetc(in)) != EOF) {
+        if (is_whitespace(c)) {
+            if (count <= 0)
+                return (sigil_err_t)ERR_PDF_CONT;
+            if (ungetc(c, in) != c)
+                return (sigil_err_t)ERR_IO;
+
+            if (strncmp(tmp, "xref", 4) == 0) {
+                *keyword = KEYWORD_xref;
+                return (sigil_err_t)ERR_NO;
+            }
+            if (strncmp(tmp, "trailer", 7) == 0) {
+                *keyword = KEYWORD_trailer;
+                return (sigil_err_t)ERR_NO;
+            }
+            return (sigil_err_t)ERR_PDF_CONT;
+        } else {
+            if (count >= keyword_max - 1)
+                return (sigil_err_t)ERR_PDF_CONT;
+            tmp[count] = c;
+            count++;
+        }
+    }
+
+    return (sigil_err_t)ERR_PDF_CONT;
+}
+
+sigil_err_t parse_free_indicator(FILE *in, free_indicator_t *result)
+{
+    sigil_err_t err;
     char c;
 
-    // skip leading whitespaces
-    while ((c = fgetc(in)) != EOF && is_whitespace(c))
-        ;
+    err = skip_leading_whitespaces(in);
+    if (err != ERR_NO)
+        return err;
+
+    c = fgetc(in);
 
     switch(c) {
         case 'f':
+            *result = FREE_ENTRY;
+            return (sigil_err_t)ERR_NO;
         case 'n':
-            *result = c;
-            return 0;
+            *result = IN_USE_ENTRY;
+            return (sigil_err_t)ERR_NO;
         default:
-            return 1;
+            return (sigil_err_t)ERR_PDF_CONT;
     }
 }
 
@@ -97,7 +166,7 @@ void print_test_item(const char *test_name, int verbosity)
     if (verbosity < 2)
         return;
 
-    printf("    - %-30s", test_name);
+    printf("    - %-32s", test_name);
 }
 
 void print_test_result(int result, int verbosity)
@@ -177,6 +246,138 @@ int sigil_auxiliary_self_test(int verbosity)
 
     print_test_result(1, verbosity);
 
+    // TEST: fn skip_leading_whitespaces
+    print_test_item("fn skip_leading_whitespaces", verbosity);
+
+    {
+        char *sstream_1 = "x";
+        char *sstream_2 = "\x00\x09\x0a\x0c\x0d\x20x";
+        FILE *file;
+
+        file = fmemopen(sstream_1,
+                        (strlen(sstream_1) + 1) * sizeof(*sstream_1),
+                        "r");
+        if (file == NULL)
+            goto failed;
+
+        if (skip_leading_whitespaces(file) != ERR_NO ||
+            fgetc(file) != 'x')
+        {
+            goto failed;
+        }
+
+        fclose(file);
+
+        file = fmemopen(sstream_2,
+                        (7 + 1) * sizeof(*sstream_2), // cannot use strlen
+                        "r");
+        if (file == NULL)
+            goto failed;
+
+        if (skip_leading_whitespaces(file) != ERR_NO ||
+            fgetc(file) != 'x')
+        {
+            goto failed;
+        }
+
+        fclose(file);
+    }
+
+    print_test_result(1, verbosity);
+
+    // TEST: fn parse_number
+    print_test_item("fn parse_number", verbosity);
+
+    {
+        char *sstream = "0123456789    42";
+        size_t result = 0;
+        FILE *file;
+
+        file = fmemopen(sstream,
+                        (strlen(sstream) + 1) * sizeof(*sstream),
+                        "r");
+        if (file == NULL)
+            goto failed;
+
+        if (parse_number(file, &result) != ERR_NO ||
+            result != 123456789)
+        {
+            goto failed;
+        }
+
+        if (parse_number(file, &result) != ERR_NO ||
+            result != 42)
+        {
+            goto failed;
+        }
+
+        fclose(file);
+    }
+
+    print_test_result(1, verbosity);
+
+    // TEST: fn parse_keyword
+    print_test_item("fn parse_keyword", verbosity);
+
+    {
+        char *sstream = " xref \n trailer";
+        keyword_t result;
+        FILE *file;
+
+        file = fmemopen(sstream,
+                        (strlen(sstream) + 1) * sizeof(*sstream),
+                        "r");
+        if (file == NULL)
+            goto failed;
+
+        if (parse_keyword(file, &result) != ERR_NO ||
+            result != KEYWORD_xref)
+        {
+            goto failed;
+        }
+
+        if (parse_keyword(file, &result) != ERR_NO ||
+            result != KEYWORD_trailer)
+        {
+            goto failed;
+        }
+
+        fclose(file);
+    }
+
+    print_test_result(1, verbosity);
+
+    // TEST: fn parse_free_indicator
+    print_test_item("fn parse_free_indicator", verbosity);
+
+    {
+        char *sstream = " f n";
+        free_indicator_t result;
+        FILE *file;
+
+        file = fmemopen(sstream,
+                        (strlen(sstream) + 1) * sizeof(*sstream),
+                        "r");
+        if (file == NULL)
+            goto failed;
+
+        if (parse_free_indicator(file, &result) != ERR_NO ||
+            result != FREE_ENTRY)
+        {
+            goto failed;
+        }
+
+        if (parse_free_indicator(file, &result) != ERR_NO ||
+            result != IN_USE_ENTRY)
+        {
+            goto failed;
+        }
+
+        fclose(file);
+    }
+
+    print_test_result(1, verbosity);
+
     // all tests done
     print_module_result(1, verbosity);
     return 0;

lib/error.c

@@ -4,7 +4,7 @@
 #include "error.h"
 
 
-const char_t *sigil_err_string(sigil_err_t err)
+const char *sigil_err_string(sigil_err_t err)
 {
     if (err == ERR_NO)
         return "NO ERROR";

lib/header.c

@@ -81,19 +81,18 @@ int sigil_header_self_test(int verbosity)
 {
     print_module_name("header", verbosity);
 
-    // prepare
     sigil_t *sgl = NULL;
 
     if (sigil_init(&sgl) != ERR_NO)
         goto failed;
 
     // TEST: correct_1
-    print_test_item("correct_1", verbosity);
+    print_test_item("fn process_header", verbosity);
 
-    char_t *correct_1 = "\x25PDF-1.1\n"             \
-                        "abcdefghijklmnopqrstuvwxyz";
-    sgl->file = fmemopen(correct_1,
-                         (strlen(correct_1) + 1) * sizeof(*correct_1),
+    char *sstream_1 = "\x25PDF-1.1\n"             \
+                      "abcdefghijklmnopqrstuvwxyz";
+    sgl->file = fmemopen(sstream_1,
+                         (strlen(sstream_1) + 1) * sizeof(*sstream_1),
                          "r");
     if (sgl->file == NULL)
         goto failed;
@@ -106,25 +105,20 @@ int sigil_header_self_test(int verbosity)
         goto failed;
     }
 
-    print_test_result(1, verbosity);
-
     fclose(sgl->file);
     sgl->file = NULL;
 
-    // TEST: correct_2
-    print_test_item("correct_2", verbosity);
-
-    char_t *correct_2 = "\x1a\x5e\x93\x7e\x6f\x3c\x6a\x71\xbf\xda\x54\x91\xe5" \
-                        "\x86\x08\x84\xaf\x8e\x89\x44\xab\xc4\x58\x0c\xb9\x31" \
-                        "\xd3\x8c\x0f\xc0\x43\x1a\xa5\x07\x4f\xe2\x98\xb3\xd8" \
-                        "\x53\x4b\x5d\x4b\xd6\x48\x26\x98\x09\xde\x0d"         \
-                        "\x25PDF-1.2"                                          \
-                        "\x55\xa1\x77\xd3\x47\xab\xc6\x87\xf3\xbc\x2d\x8a\x9f" \
-                        "\x0e\x47\xbb\x74\xd2\x71\x28\x94\x53\x92\xae\x2b\x17" \
-                        "\xd0\x6a\x9c\x13\x84\xc1\x07\x44\xc0\x81\xb8\xd6\x9c" \
-                        "\x31\x08\x13\xd4\xc2\xd6\x2d\xaf\xfb\xea\x6f";
-    sgl->file = fmemopen(correct_2,
-                         (strlen(correct_2) + 1) * sizeof(*correct_2),
+    char *sstream_2 = "\x1a\x5e\x93\x7e\x6f\x3c\x6a\x71\xbf\xda\x54\x91\xe5"\
+                      "\x86\x08\x84\xaf\x8e\x89\x44\xab\xc4\x58\x0c\xb9\x31"\
+                      "\xd3\x8c\x0f\xc0\x43\x1a\xa5\x07\x4f\xe2\x98\xb3\xd8"\
+                      "\x53\x4b\x5d\x4b\xd6\x48\x26\x98\x09\xde\x0d"        \
+                      "\x25PDF-1.2"                                         \
+                      "\x55\xa1\x77\xd3\x47\xab\xc6\x87\xf3\xbc\x2d\x8a\x9f"\
+                      "\x0e\x47\xbb\x74\xd2\x71\x28\x94\x53\x92\xae\x2b\x17"\
+                      "\xd0\x6a\x9c\x13\x84\xc1\x07\x44\xc0\x81\xb8\xd6\x9c"\
+                      "\x31\x08\x13\xd4\xc2\xd6\x2d\xaf\xfb\xea\x6f";
+    sgl->file = fmemopen(sstream_2,
+                         (strlen(sstream_2) + 1) * sizeof(*sstream_2),
                          "r");
     if (sgl->file == NULL)
         goto failed;
@@ -137,17 +131,12 @@ int sigil_header_self_test(int verbosity)
         goto failed;
     }
 
-    print_test_result(1, verbosity);
-
     fclose(sgl->file);
     sgl->file = NULL;
 
-    // TEST: wrong_1
-    print_test_item("wrong_1", verbosity);
-
-    char_t *wrong_1 = "\x25\x25PPD\x25PDF-\x25PDF-1\x25PDF-1..@PDF-1.3";
-    sgl->file = fmemopen(wrong_1,
-                         (strlen(wrong_1) + 1) * sizeof(*wrong_1),
+    char *sstream_3 = "\x25\x25PPD\x25PDF-\x25PDF-1\x25PDF-1..@PDF-1.3";
+    sgl->file = fmemopen(sstream_3,
+                         (strlen(sstream_3) + 1) * sizeof(*sstream_3),
                          "r");
     if (sgl->file == NULL)
         goto failed;
@@ -155,18 +144,17 @@ int sigil_header_self_test(int verbosity)
     if (process_header(sgl) == ERR_NO)
         goto failed;
 
-    print_test_result(1, verbosity);
+    sigil_free(sgl);
 
-    fclose(sgl->file);
-    sgl->file = NULL;
+    print_test_result(1, verbosity);
 
     // all tests done
     print_module_result(1, verbosity);
     return 0;
 
 failed:
-    if (sgl->file)
-        fclose(sgl->file);
+    if (sgl)
+        sigil_free(sgl);
 
     print_test_result(0, verbosity);
     print_module_result(0, verbosity);

lib/sigil.c

@@ -2,20 +2,13 @@
 #include <stdio.h>
 #include <string.h>
 #include "auxiliary.h"
+#include "error.h"
 #include "header.h"
 #include "sigil.h"
 #include "trailer.h"
 #include "xref.h"
 
 
-static sigil_err_t validate_mode(mode_t mode)
-{
-    if (mode != MODE_VERIFY && mode != MODE_SIGN)
-        return (sigil_err_t)ERR_PARAM;
-
-    return (sigil_err_t)ERR_NO;
-}
-
 sigil_err_t sigil_init(sigil_t **sgl)
 {
     // function parameter checks
@@ -30,7 +23,6 @@ sigil_err_t sigil_init(sigil_t **sgl)
     // set default values
     (*sgl)->file             = NULL;
     (*sgl)->filepath         = NULL;
-    (*sgl)->mode             = MODE_UNSET;
     (*sgl)->pdf_x            = 0;
     (*sgl)->pdf_y            = 0;
     (*sgl)->xref_type        = XREF_TYPE_UNSET;
@@ -42,7 +34,7 @@ sigil_err_t sigil_init(sigil_t **sgl)
     return (sigil_err_t)ERR_NO;
 }
 
-sigil_err_t sigil_setup_file(sigil_t *sgl, const char_t *filepath)
+sigil_err_t sigil_setup_file(sigil_t *sgl, const char *filepath)
 {
     // function parameter checks
     if (sgl == NULL || filepath == NULL)
@@ -66,36 +58,16 @@ sigil_err_t sigil_setup_file(sigil_t *sgl, const char_t *filepath)
     return (sigil_err_t)ERR_NO;
 }
 
-sigil_err_t sigil_setup_mode(sigil_t *sgl, mode_t mode)
-{
-    // function parameter checks
-    if (sgl == NULL || validate_mode(mode) != ERR_NO)
-        return (sigil_err_t)ERR_PARAM;
-
-    sgl->mode = mode;
-
-    return (sigil_err_t)ERR_NO;
-}
-
-sigil_err_t sigil_process(sigil_t *sgl)
+sigil_err_t sigil_verify(sigil_t *sgl)
 {
     sigil_err_t err;
 
     // function parameter checks