[Refdb-cvs] CVS: refdb/src authorinfo.c,1.3,1.4 tokenize.c,1.10,1.11

[Markus H. <mho...@us...>]

Update of /cvsroot/refdb/refdb/src
In directory sc8-pr-cvs1:/tmp/cvs-serv722

Modified Files:
	authorinfo.c tokenize.c 
Log Message:
added support for hyphenated double initials in author names

Index: authorinfo.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/authorinfo.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -U2 -r1.3 -r1.4
--- authorinfo.c	28 Dec 2003 00:43:19 -0000	1.3
+++ authorinfo.c	6 Jan 2004 23:06:36 -0000	1.4
@@ -45,4 +45,5 @@
   char* temp_middle;
   char* new_temp_author;
+  char bitsandpieces[6];
   size_t temp_author_len = 256;
 
@@ -71,15 +72,35 @@
     }
 
-    if ((new_temp_author = mstrcat(temp_author, ptr_ainfo->firstname, &temp_author_len, 0)) == NULL) {
-      (*ptr_error)++;
-      free(temp_author);
-      return NULL;
+    /* we've got to analyze whether this is a full first name or some
+       hyphenated double initial */
+    if (ptr_ainfo->firstname[1] == '-') {
+      *bitsandpieces = *(ptr_ainfo->firstname);
+      *(bitsandpieces + 1) = '.';
+      *(bitsandpieces + 2) = '-';
+      *(bitsandpieces + 3) = *(ptr_ainfo->firstname + 2);
+      *(bitsandpieces + 4) = '.';
+      *(bitsandpieces + 5) = '\0';
+    }
+    else if (*(ptr_ainfo->firstname + 1) == '\0') {
+      *bitsandpieces = *ptr_ainfo->firstname;
+      *(bitsandpieces + 1) = '.';
+      *(bitsandpieces + 2) = '\0';
     }
     else {
-      temp_author = new_temp_author;
+      *bitsandpieces = '\0';
     }
 
-    if (ptr_ainfo->firstname[1] == '\0') {
-      if ((new_temp_author = mstrcat(temp_author, ".", &temp_author_len, 0)) == NULL) {
+    if (*bitsandpieces) {
+      if ((new_temp_author = mstrcat(temp_author, bitsandpieces, &temp_author_len, 0)) == NULL) {
+	(*ptr_error)++;
+	free(temp_author);
+	return NULL;
+      }
+      else {
+	temp_author = new_temp_author;
+      }
+    }
+    else {
+      if ((new_temp_author = mstrcat(temp_author, ptr_ainfo->firstname, &temp_author_len, 0)) == NULL) {
 	(*ptr_error)++;
 	free(temp_author);
@@ -112,16 +133,35 @@
       }
 
-      if ((new_temp_author = mstrcat(temp_author, item, &temp_author_len, 0)) == NULL) {
-	(*ptr_error)++;
-	free(temp_author);
-	free(temp_middle);
-	return NULL;
+      /* check whether middle name is hyphenated initial */
+      if (*(item + 1) == '-') {
+	*bitsandpieces = *item;
+	*(bitsandpieces + 1) = '.';
+	*(bitsandpieces + 2) = '-';
+	*(bitsandpieces + 3) = *(item + 2);
+	*(bitsandpieces + 4) = '.';
+	*(bitsandpieces + 5) = '\0';
+      }
+      else if (*(item + 1) == '\0') {
+	*bitsandpieces = *item;
+	*(bitsandpieces + 1) = '.';
+	*(bitsandpieces + 2) = '\0';
       }
       else {
-	temp_author = new_temp_author;
+	*bitsandpieces = '\0';
       }
 
-      if (item[1] == '\0') {
-	if ((new_temp_author = mstrcat(temp_author, ".", &temp_author_len, 0)) == NULL) {
+      if (*bitsandpieces) {
+	if ((new_temp_author = mstrcat(temp_author, bitsandpieces, &temp_author_len, 0)) == NULL) {
+	  (*ptr_error)++;
+	  free(temp_author);
+	  free(temp_middle);
+	  return NULL;
+	}
+	else {
+	  temp_author = new_temp_author;
+	}
+      }
+      else {
+	if ((new_temp_author = mstrcat(temp_author, item, &temp_author_len, 0)) == NULL) {
 	  (*ptr_error)++;
 	  free(temp_author);

Index: tokenize.c
===================================================================
RCS file: /cvsroot/refdb/refdb/src/tokenize.c,v
retrieving revision 1.10
retrieving revision 1.11
diff -u -U2 -r1.10 -r1.11
--- tokenize.c	6 Jan 2004 15:41:57 -0000	1.10
+++ tokenize.c	6 Jan 2004 23:06:36 -0000	1.11
@@ -971,11 +971,33 @@
       }
 
-      /* now there's two options: 
+      /* now there's three options: 
 	 - the next char is the first initial, followed by a period 
+	 - the next char is the first initial of a hyphenated double
+	   name
 	 - the next word is the full first name */
       
       if (*(ptr_atoken->first + 1) == '.') {
-	*(ptr_atoken->first + 1) = '\0';
-	middle = ptr_atoken->first + 2;
+	if (*(ptr_atoken->first + 2) == '-') {
+	  /* now we've got to find the end of the hyphenated name, remove
+	     the periods, and terminate the string appropriately */
+	  middle = ptr_atoken->first + 3;
+	  while (*middle && *middle != ' ') {
+	    middle++;
+	  }
+
+	  if (*(middle-1) == '.') {
+	    *(middle-1) = '\0';
+	  }
+	  else if (*middle == ' ') {
+	    *middle = '\0';
+	    middle++;
+	  }
+	  
+	  memmove(ptr_atoken->first + 1, ptr_atoken->first + 2, strlen(ptr_atoken->first + 2)+1);
+	}
+	else {
+	  *(ptr_atoken->first + 1) = '\0';
+	  middle = ptr_atoken->first + 2;
+	}
       }
       else if (*(ptr_atoken->first + 1)) { /* string doesn't end here */
@@ -1008,6 +1030,26 @@
 	  /* middle name could be abbreviated, so look for a dot */
 	  if (*(middle + 1) == '.') {
-	    *(middle + 1) = '\0';
-	    next_middle = middle + 2;
+	    if (*(middle + 2) == '-') {
+	      /* now we've got to find the end of the hyphenated name, remove
+		 the periods, and terminate the string appropriately */
+	      next_middle = middle + 3;
+	      while (*next_middle && *next_middle != ' ') {
+		next_middle++;
+	      }
+
+	      if (*(next_middle-1) == '.') {
+		*(next_middle-1) = '\0';
+	      }
+	      else if (*next_middle == ' ') {
+		*next_middle = '\0';
+		next_middle++;
+	      }
+	  
+	      memmove(middle + 1, middle + 2, strlen(middle + 2)+1);
+	    }
+	    else {
+	      *(middle + 1) = '\0';
+	      next_middle = middle + 2;
+	    }
 	  }
 	  else {