Add support for annotation authors

Author: semvis123

catdoc.go

@@ -102,6 +102,14 @@ func GetCommentsFromFile(file io.ReadSeeker) (string, error) {
 	return callWASMFuncWithFile("get_comments", file)
 }
 
+func GetAnnotationAuthorsFromFile(file io.ReadSeeker) ([]string, error) {
+	r, err := callWASMFuncWithFile("get_annotation_authors", file)
+	if err != nil {
+		return nil, err
+	}
+	return strings.Split(r, "\n"), nil
+}
+
 func GetVersion() (string, error) {
 	return callWASMFunc("get_version", nil)
 }

catdoc.wasm

@@ -105,7 +105,7 @@ catdoc-wasm: emsdk.uptodate
 		-I. -O3 -DHAVE_CONFIG_H \
 		-DCATDOC_VERSION=\"0.95\" \
 		-DCHARSETPATH=\"charsets\" \
-		-sEXPORTED_FUNCTIONS=_get_text,_get_author,_get_last_author,_get_version,_get_title,_get_subject,_get_keywords,_get_comments\
+		-sEXPORTED_FUNCTIONS=_get_text,_get_author,_get_last_author,_get_version,_get_title,_get_subject,_get_keywords,_get_comments,_get_annotation_authors\
 		-sSTANDALONE_WASM -sWARN_ON_UNDEFINED_SYMBOLS=0 \
 		--no-entry -sFILESYSTEM=1 -sALLOW_MEMORY_GROWTH -sMAXIMUM_MEMORY=1GB
 

catdoc/src/Makefile

@@ -18,6 +18,9 @@ extern char ole_sign[], zip_sign[]; /* from ole.c */
 char rtf_sign[] = "{\\rtf";
 char old_word_sign[] = {0xdb, 0xa5, 0};
 char write_sign[] = {0x31, 0xBE, 0};
+unsigned char sttbfRMark[] = {0x00, 0x00, 0x07, 0x00, 0x55, 0x00,
+                              0x6E, 0x00, 0x6B, 0x00, 0x6E, 0x00,
+                              0x6F, 0x00, 0x77, 0x00, 0x6E, 0x00};
 int verbose = 0;
 
 unsigned char *read_metadata(unsigned char *buffer, metadata metadata_type) {
@@ -38,13 +41,75 @@ unsigned char *read_metadata(unsigned char *buffer, metadata metadata_type) {
   return NULL;
 }
 
+int find_offset(FILE *f, unsigned char *mark, int mark_length) {
+  int offset = 0;
+  int block_size = 1024 * 25;
+  int pos = 0;
+  int bpos = 0;
+  unsigned char buf[block_size];
+  int block_n = 0;
+
+  while (!feof(f)) {
+    long n = fread(buf, 1, block_size, f);
+    while (bpos != n) {
+      if (buf[bpos] == mark[pos]) {
+        pos++;
+        bpos++;
+        if (pos == mark_length) {
+          offset = block_n * block_size + bpos - pos;
+          return offset;
+        }
+      } else {
+        bpos -= pos;
+        bpos++;
+        pos = 0;
+      }
+    }
+    bpos = 0;
+    block_n++;
+  }
+  fprintf(stderr, "stttbfRMark offset is not found");
+  exit(1);
+}
+
+void read_annotation_authors(FILE *f) {
+  int offset = find_offset(f, sttbfRMark, 18) - 4;
+  fseek(f, offset, SEEK_SET);
+
+  int block_size = 1024;
+  unsigned char buf[2];
+  fread(buf, 1, 2, f);
+  if (buf[0] != 0xff || buf[1] != 0xff) {
+    fprintf(stderr, "stttbfRMark offset is invalid");
+    exit(1);
+  }
+  fread(buf, 1, 2, f);
+  unsigned int count = getshort(buf, 0) - 1;
+  fseek(f, 18, SEEK_CUR);
+  for (int i = 0; i < count; i++) {
+    fread(buf, 1, 2, f);
+    unsigned int str_len = getshort(buf, 0);
+    unsigned short *str = calloc(str_len, 2);
+    fread(str, 2, str_len, f);
+    for (int j = 0; j < str_len; j++) {
+      printf("%lc", str[j]);
+    }
+    printf("\n");
+    free(str);
+  }
+}
+
 /*********************************************************************
  * Determines format of input file and calls parse_word_header or
  * process_file if
  * it is word processor file or copy_out if it is plain text file
  * return not 0 when error
  ********************************************************************/
 int analyze_format(FILE *f, metadata metadata_type) {
+  if (metadata_type == annotation_authors) {
+    read_annotation_authors(f);
+    return 0;
+  }
   unsigned char buffer[129];
   long offset = 0;
   FILE *new_file, *ole_file;

catdoc/src/analyze.c

@@ -56,7 +56,7 @@ int main(int argc, char **argv) {
   get_locale_charset();
 #endif
   metadata metadata_type = none;
-  while ((c = getopt(argc, argv, "Vls:d:f:taubxv8wALTSKCm:")) != -1) {
+  while ((c = getopt(argc, argv, "Vls:d:f:taubxv8wALTSKCUm:")) != -1) {
     switch (c) {
     case 's':
       check_charset(&source_csname, optarg);
@@ -104,6 +104,9 @@ int main(int argc, char **argv) {
     case 'C':
       metadata_type = comments;
       break;
+    case 'U':
+      metadata_type = annotation_authors;
+      break;
     case 'm': {
       char *endptr;
       wrap_margin = (int)strtol(optarg, &endptr, 0);
@@ -246,3 +249,8 @@ void get_comments() {
   char *args[] = {"", "-C", "/input_file/file.doc"};
   main(3, args);
 }
+
+void get_annotation_authors() {
+  char *args[] = {"", "-U", "/input_file/file.doc"};
+  main(3, args);
+}

catdoc/src/catdoc.c

@@ -227,6 +227,7 @@ void get_title();
 void get_subject();
 void get_keywords();
 void get_comments();
+void get_annotation_authors();
 
 char *find_file(char *name, const char *path);
 char *stradd(const char *s1, const char *s2);
@@ -255,7 +256,8 @@ typedef enum {
   title,
   subject,
   keywords,
-  comments
+  comments,
+  annotation_authors
 } metadata;
 int analyze_format(FILE *f, metadata metadata_type);
 void list_charsets(void);

catdoc/src/catdoc.h

@@ -31,6 +31,22 @@ func testFileFunc(title, expected string, fun func(io.ReadSeeker) (string, error
 	}
 }
 
+func testFileFuncArr(title string, expected []string, fun func(io.ReadSeeker) ([]string, error), t *testing.T) {
+	f, err := os.Open("test.doc")
+	if err != nil {
+		t.Fatalf("could not open test document, %v", err)
+	}
+	arr, err := fun(f)
+	if err != nil {
+		t.Fatalf("expected error to be nil, got %v", err)
+	}
+	for i := range expected {
+		if arr[i] != expected[i] {
+			t.Fatalf("expected %s to be \"%v\", got %v", title, expected, arr)
+		}
+	}
+}
+
 func TestGetTextFromFile(t *testing.T) {
 	testFileFunc("text", "text-inside-doc", gocatdoc.GetTextFromFile, t)
 }
@@ -58,3 +74,7 @@ func TestGetKeywordsFromFile(t *testing.T) {
 func TestGetCommentsFromFile(t *testing.T) {
 	testFileFunc("comments", "Comments", gocatdoc.GetCommentsFromFile, t)
 }
+
+func TestGetAnnotationAuthorsFromFile(t *testing.T) {
+	testFileFuncArr("annoation_authors", []string{"H. Potter"}, gocatdoc.GetAnnotationAuthorsFromFile, t)
+}