some more compiler attributes

make sure all textcode/scan_* routines work if dest or destlen is NULL
This commit is contained in:
leitner
2025-03-17 13:21:55 +00:00
parent 144912a41d
commit fb5dcdf7fe
6 changed files with 56 additions and 43 deletions

View File

@@ -191,11 +191,18 @@
// __strout: Argument idx (counted left to right from 1) is a non-NULL string the function writes
// Currently we can't tell the compiler that the output will be 0-terminated.
#define __strout(idx) __attribute__((nonnull(idx),__access(write_only,idx)))
// __strout_ornull: Argument idx (counted left to right from 1) is a string the function writes (may be NULL)
// Currently we can't tell the compiler that the output will be 0-terminated.
#define __strout_ornull(idx) __attribute__((__access(write_only,idx)))
// __strnout: Argument idx (counted left to right from 1) is a non-NULL string the function writes, buffer size given in szidx'th function argument (in array elements, not necessarily bytes)
// Currently we can't tell the compiler that the output will be 0-terminated.
#define __strnout(idx,szidx) __attribute__((nonnull(idx),__access3(write_only,idx,szidx)))
// __bufout: Argument idx (counted left to right from 1) is a non-NULL buffer the function writes, buffer size given in szidx'th function argument (in array elements, not necessarily bytes)
#define __bufout(bufidx,szidx) __attribute__((nonnull(bufidx),__access3(write_only,bufidx,szidx)))
// __retarg: Returns value in argument idx (counted left to right from 1) which must be non-NULL
#define __retarg(bufidx) __attribute__((nonnull(bufidx),__access(write_only,bufidx)))
// __retarg_ornull: Returns value in argument idx (counted left to right from 1) which can be NULL
#define __retarg_ornull(bufidx) __attribute__((__access(write_only,bufidx)))
// __strinout: Argument idx (counted left to right from 1) is a non-NULL string the function reads and writes
// Currently we can't tell the compiler that the output will be 0-terminated.
#define __strinout(idx) __attribute__((nonnull(idx),__access(read_write,idx),null_terminated_string_arg(idx)))

View File

@@ -3,6 +3,7 @@
#define TEXTCODE_H
#include <stddef.h>
#include <libowfat/compiler.h>
#ifdef __cplusplus
extern "C" {
@@ -11,120 +12,125 @@ extern "C" {
/* These take len bytes from src and write them in encoded form to
* dest (if dest != NULL), returning the number of bytes written. */
/* dietlibc defines these in sys/cdefs.h, which is included from stddef.h */
#ifndef __readmemsz__
#define __readmemsz__(a,b)
#endif
#ifndef __writemem__
#define __writemem__(a)
#endif
/* Needs len/3*4 bytes */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_uuencoded(char* dest,const char* src,size_t len);
/* Needs len/3*4 bytes */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_base64(char* dest,const char* src,size_t len);
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_base64url(char* dest,const char* src,size_t len);
/* Worst case: len*3 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_quotedprintable(char* dest,const char* src,size_t len);
/* Worst case: len*3 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_quotedprintable2(char* dest,const char* src,size_t len,const char* escapeme);
/* Worst case: len*3 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_urlencoded(char* dest,const char* src,size_t len);
/* Worst case: len*3 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_urlencoded2(char* dest,const char* src,size_t len,const char* escapeme);
/* Worst case: len*2 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_yenc(char* dest,const char* src,size_t len);
/* Needs len*2 bytes */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_hexdump(char* dest,const char* src,size_t len);
/* Change '<' to '&lt;' and '&' to '&amp;' and '\n' to '<br>'; worst case: len*5 */
/* This is meant for outputting text that goes between tags */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_html(char* dest,const char* src,size_t len);
/* Change '<' to '&lt;' and '&' to '&amp;' and '"' to '&quot;'; worst case: len*6 */
/* This is meant for outputting text that goes in a tag argument between double quotes*/
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_html_tagarg(char* dest,const char* src,size_t len);
/* Change '<' to '&lt;' and '&' to '&amp;'; worst case: len*5 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_xml(char* dest,const char* src,size_t len);
/* Change '\' to "\\", '\n' to "\n", ^A to "\x01" etc; worst case: len*4 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_cescape(char* dest,const char* src,size_t len);
/* Worst case: len*4 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_cescape2(char* dest,const char* src,size_t len,const char* escapeme);
/* Fold whitespace to '_'; this is great for writing fields with
* white spaces to a log file and still allow awk to do log analysis */
/* Worst case: same size */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_foldwhitespace(char* dest,const char* src,size_t len);
/* Worst case: len*3 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_ldapescape(char* dest,const char* src,size_t len);
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_ldapescape2(char* dest,const char* src,size_t len,const char* escapeme);
/* Encode JSON string from UTF-8; will backslash-escape the bare minimum.
* Invalid UTF-8 in input will output as valid UTF-8 for each byte
* Worst case: len*6 */
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_jsonescape(char* dest,const char* src,size_t len);
__writemem__(1) __readmemsz__(2,3)
__strout_ornull(1) __bufin(2,3)
size_t fmt_base85(char* dest,const char* src,size_t len);
/* These read one line from src, decode it, and write the result to
* dest. The number of decoded bytes is written to destlen. dest
* should be able to hold strlen(src) bytes as a rule of thumb. */
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_uuencoded(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_base64(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_base64url(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_quotedprintable(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_urlencoded(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_urlencoded2(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_yenc(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_hexdump(const char* src,char* dest,size_t* destlen);
/* decodes all html5-standardized &foo; escapes, and also
* "<br>" to "\n" and "<p>" to "\n\n", leaves the rest of the tags alone */
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_html(const char* src,char* dest,size_t* destlen);
/* decodes all html5-standardized &foo; escapes, but leaves all tags
* alone */
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_html_tagarg(const char* src,char* dest,size_t* destlen);
/* Decodes escaped C string text, turning e.g. \n into newline */
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_cescape(const char* src,char* dest,size_t* destlen);
/* Decodes escaped LDIF text, turning e.g. \5C into \ */
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_ldapescape(const char* src,char* dest,size_t* destlen);
/* Decodes escaped JSON strings (like \"\r\n\u0013), turns escaped
* surrogate pairs into UTF-8. Expects input to be valid UTF-8. Ends at
* \0 or unescaped double quote. */
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_jsonescape(const char* src,char* dest,size_t* destlen);
__strin(1) __strout_ornull(2) __retarg_ornull(3)
size_t scan_base85(const char* src,char* dest,size_t* destlen);
#ifdef STRALLOC_H

View File

@@ -63,11 +63,11 @@ size_t scan_jsonescape(const char *src,char *dest,size_t *destlen) {
++written;
}
done:
*destlen=written;
if (destlen) *destlen=written;
return i;
abort:
if (prev!=(unsigned int)-1) i-=6; // if we abort and there still was an open surrogate pair, cancel it
*destlen=written;
if (destlen) *destlen=written;
return i;
}

View File

@@ -8,16 +8,16 @@ size_t scan_quotedprintable(const char *src,char *dest,size_t *destlen) {
if (s[i]=='=') {
int j=scan_fromhex(s[i+1]);
if (j<0) break;
dest[written]=j<<4;
if (dest) dest[written]=j<<4;
j=scan_fromhex(s[i+2]);
if (j<0) break;
dest[written]|=j;
if (dest) dest[written]|=j;
i+=2;
} else {
dest[written]=s[i];
if (dest) dest[written]=s[i];
}
++written;
}
*destlen=written;
if (destlen) *destlen=written;
return i;
}

View File

@@ -12,17 +12,17 @@ static size_t inner_scan_urlencoded(const char *src,char *dest,size_t *destlen,i
c=j<<4;
j=scan_fromhex(s[i+2]);
if (j<0) break;
dest[written]=c|j;
if (dest) dest[written]=c|j;
i+=2;
} else if (s[i]=='+' && plus)
dest[written]=' ';
else if (s[i]<=' ' || s[i]=='?' || s[i]=='&')
} else if (s[i]=='+' && plus) {
if (dest) dest[written]=' ';
} else if (s[i]<=' ' || s[i]=='?' || s[i]=='&')
break; /* invalid input */
else
dest[written]=s[i];
if (dest) dest[written]=s[i];
++written;
}
*destlen=written;
if (destlen) *destlen=written;
return i;
}

View File

@@ -8,13 +8,13 @@ size_t scan_yenc(const char *src,char *dest,size_t *destlen) {
if (s[i]=='=') {
++i;
if (s[i]=='y') break;
dest[written]=s[i]-64-42;
if (dest) dest[written]=s[i]-64-42;
} else if (s[i]=='\n' || s[i]=='\r' || s[i]=='\0')
break;
else
dest[written]=s[i]-42;
if (dest) dest[written]=s[i]-42;
++written;
}
*destlen=written;
if (destlen) *destlen=written;
return i;
}