commit version 0.1

This commit is contained in:
2017-08-08 22:24:40 +02:00
parent e3d663d6c3
commit 5378676596
17 changed files with 995 additions and 1 deletions
+9
View File
@@ -0,0 +1,9 @@
set(pwd ${CMAKE_CURRENT_SOURCE_DIR})
set(SOURCE
${SOURCE}
${pwd}/main.c
${pwd}/nhtml_string.c
${pwd}/attribute.c
${pwd}/html.c
${pwd}/includes.c
PARENT_SCOPE)
+66
View File
@@ -0,0 +1,66 @@
/*
* attribute.c
*
* Created on: 08.08.2017
* Author: julian
*/
#include "attribute.h"
#include <string.h> // needs: strcmp
#include <stdlib.h> // needs: realloc
// linear search in set
// returns NULL on failure to find entry
attr_t * attr_set_find(attr_set_t *set, const char * name) {
size_t current = 0;
for (; current != set->len; ++current) {
if (strcmp(set->arr[current].name.c_str, name) == 0) {
return set->arr + current;
}
}
return NULL;
}
// Append new Attribute to set
int attr_set_append(attr_set_t * set, attr_t *new_entry) {
if (new_entry->name.c_str== NULL) return -1; // reject empty entries
// search first
attr_t * new_ptr = attr_set_find(set, new_entry->name.c_str);
if (new_ptr != NULL) {
// already contained in set
// concatenate values
//new_ptr->value = string_copy(new_entry->value);
string_append(&new_ptr->value, ' ');
string_concat(&new_ptr->value, &new_entry->value);
return 0;
}
new_ptr = realloc(set->arr, (set->len+1)*sizeof(attr_t));
if (new_ptr == NULL) {
return -1;
}
// append and quit
set->arr = new_ptr;
set->arr[set->len] = attr_copy(new_entry);
set->len++;
return 0;
}
attr_t attr_copy(attr_t * attr) {
attr_t temp;
temp.name = string_copy(attr->name);
temp.value = string_copy(attr->value);
return temp;
}
void attr_destroy(attr_t * attr) {
// clear memory
string_destroy(&attr->name);
string_destroy(&attr->value);
}
+71
View File
@@ -0,0 +1,71 @@
/*
* html.c
*
* Created on: 07.08.2017
* Author: julian
*/
#include "html.h"
int html_escape(int c, FILE * output) {
switch(c) {
case '<':
fprintf(output, "&lt;");
goto escaped;
case '>':
fprintf(output, "&gt;");
goto escaped;
case '&':
fprintf(output, "&amp;");
goto escaped;
case '"':
fprintf(output, "&quot;");
goto escaped;
/*case ' ':
fprintf(output, "&nbsp;");
goto escaped;*/
case '\n':
fprintf(output, "<br/>");
goto escaped;
}
return 0;
escaped:
return 1;
}
// write out node
void open_node(node_t * node, FILE * output) {
// check for empty node (text mostly)
if (!node->name.c_str) return;
// start by writing tag
fputc('<', output);
// follow with tag name
fputs(node->name.c_str,output);
// add attributes
size_t i = 0;
attr_t * current = node->attributes.arr;
for (; i < node->attributes.len; i++,current++) {
if (current->value.c_str)
fprintf(output, " %s=\"%s\"", current->name.c_str, current->value.c_str);
else {
fputc(' ', output);
fputs(current->name.c_str, output);
}
}
// close tag
fputc('>', output);
}
void close_node(node_t * node, FILE * output) {
if (node->name.c_str) {
fprintf(output, "</%s>", node->name.c_str);
}
}
+94
View File
@@ -0,0 +1,94 @@
/*
* includes.c
*
* Created on: 08.08.2017
* Author: julian
*/
#include <includes.h>
#include <parser.h>
#include <nhtml_string.h>
#include <stdlib.h> // needs: malloc, realloc
#include <memory.h> // needs: memcpy
string_t* include_paths;
size_t num_includes = 0;
void include_paths_init() {
include_paths = malloc(sizeof(string_t));
num_includes = 1;
include_paths[0] = string_from_cstr("."); // search in current dir by default
}
const char * extension(const string_t * path) {
int i;
for (i = path->len; i > 0; i--) {
if (path->c_str[i] == '/')
return NULL;
if (path->c_str[i] == '.')
break;
}
return path->c_str + i;
}
void include_add_path(const char * path) {
if (num_includes == 0) {
include_paths_init();
}
include_paths = realloc(include_paths, sizeof(const char*)*(num_includes+1));
include_paths[num_includes] = string_from_cstr(path);
num_includes++;
}
int include_file(string_t *filename, FILE * output) {
FILE * input = NULL;
char * tempPath = NULL;
if(num_includes == 0) {
include_paths_init();
}
for (size_t i = 0; i < num_includes; ++i) {
// actual length is length + 2 (2* end char)
tempPath = malloc(filename->len + include_paths[i].len);
if (tempPath == NULL) {
continue;
}
// Concatenate the two paths
// adding a safety slash to separate the two
memcpy(tempPath, include_paths[i].c_str, include_paths[i].len-1);
tempPath[include_paths->len-1] = '/';
memcpy(tempPath + include_paths->len, filename->c_str, filename->len);
input = fopen(tempPath, "r");
if (input != NULL) {
break;
}
}
if (input == NULL) {
return 0;
}
const char * ext = extension(filename);
#ifdef DEBUG
printf("include %s, ext: %s\n", filename->c_str, ext);
#endif
if (ext && !strcmp(ext, ".nhtml")) {
// (re)parse whole file
int current = strip(input);
while((current = parse_node(current, input, output)) != EOF);
} else {
// copy file 1:1
int current = 0;
while((current = fgetc(input)) != EOF) {
fputc(current, output);
}
}
fclose(input);
return 1;
}
+339
View File
@@ -0,0 +1,339 @@
#include <stdio.h> // needs: fgetc, fputs, fopen, fprintf
#include <stdlib.h> // needs: abort
#include <errno.h>
#include <string.h> // needs: sterror
#include <ctype.h> // needs: isspace
#include <getopt.h> // needs: getopt_long
// Project specific includes
#include <html.h>
#include <attribute.h>
#include <nhtml_string.h>
#include <includes.h>
// globals
int verbose = 0;
int strip(FILE * stream) {
int current = 0;
while((current = fgetc(stream)) != EOF) {
if (!isspace(current))
break;
}
return current;
}
int parse_attr(FILE * stream, attr_set_t * output) {
attr_t current_attr = {};
#ifdef DEBUG
printf("parse_attr\n");
#endif
int buffer = 0;
unsigned char isKey = 1;
while((buffer = fgetc(stream)) != EOF) {
// parse key=value pairs
// check for delim
//if (buffer == ' ' || buffer == '\t' || buffer == '\n'){
if (isspace(buffer)) {
attr_set_append(output, &current_attr);
#ifdef DEBUG
printf("parsed attr: %s=%s\n", current_attr.name.c_str, current_attr.value);
#endif
// reset attribute
attr_destroy(&current_attr);
isKey = 1;
continue;
}
if (isKey && buffer == '.') {
isKey = 0;
current_attr.name = string_from_cstr("class");
continue;
}
if (buffer == '=') {
isKey = 0;
continue;
} else
if (buffer == ']') {
break;
}
if (isKey) {
string_append(&current_attr.name, buffer);
} else {
string_append(&current_attr.value, buffer);
}
}
// append last attribute
attr_set_append(output, &current_attr);
#ifdef DEBUG
printf("parsed attr: %s=%s\n", current_attr.name.c_str, current_attr.value);
#endif
attr_destroy(&current_attr);
return strip(stream);
}
int parse_text(int end_char, FILE * stream, FILE * output) {
#ifdef DEBUG
printf("parse text\n");
#endif
int buffer = 0;
char escaped = 0;
while((buffer = fgetc(stream)) != EOF) {
if (!escaped && buffer == '\\') {
escaped = 1;
continue;
}
if (!escaped && buffer == end_char) {
break;
}
escaped = 0;
if (end_char == '"' && html_escape(buffer, output)) {
continue;
}
fputc(buffer, output);
}
return strip(stream);
}
int readName(FILE *stream, node_t *node) {
int current = strip(stream);
if (current == EOF) {
return current;
}
do {
if (current == '{' || current == '[') {
break;
}
if (current == ' ' || current == '\t' || current == '\n') {
current = strip(stream);
break;
}
string_append(&node->name, current);
} while((current = fgetc(stream)) != EOF);
return current;
}
// Parse comments
int parse_comment(int current, FILE * input) {
current = fgetc(input);
if (current == '/') {
// line comment
while((current = fgetc(input)) != EOF) {
if (current == '\n') break;
}
} else
if (current == '*') {
// block comment
unsigned char comment_done = 0;
while((current = fgetc(input)) != EOF) {
if (current == '*') {
comment_done = 1;
} else
if (comment_done && current == '/') {
break;
} else
comment_done = 0;
}
}
return strip(input);
}
int parse_include(FILE * input, FILE * output) {
// handle include
int current = fgetc(input);
string_t filename = {};
// two possibilities
// 1. include with "" <= supports arbitrary paths
// 2. include without ""
// reads till a nonpath, char (!{")
// then strips empty chars from start and end of path
// both methods then try to find the file in the include paths
if (current == EOF) return current;
if (current == '"') { // Method 1
char escaped = 0;
while((current = fgetc(input)) != EOF) {
if (!escaped && current == '\\') {
escaped = 1;
continue;
}
if (!escaped && current == '"') {
break;
}
escaped = 0;
string_append(&filename, current);
}
} else {
string_append(&filename, current);
while((current = fgetc(input)) != EOF) {
switch(current) {
case '{':
case '!':
case '"':
case '}':
break;
}
if (isspace(current)) {
break;
}
string_append(&filename, current);
}
}
if (filename.c_str == NULL) {
fprintf(stderr, "include command without filename!\n");
return strip(input);
}
// handle filename
if (verbose) printf("include file %s\n", filename.c_str);
if (!include_file(&filename, output)) {
fprintf(stderr, "could not include file %s\n", filename.c_str);
}
return strip(input);
}
int parse_node(int current, FILE * stream, FILE * output) {
if (current == '"' || current == '(') {
if (current == '(') current = ')';
return parse_text(current, stream, output);
} else
if (current == '/') {
return parse_comment(current, stream);
} else
if (current == '@') {
return parse_include(stream, output);
}
// normal node
node_t current_node = {};
string_append(&current_node.name, current);
current = readName(stream, &current_node);
if (current == EOF) {
goto done;
}
#ifdef DEBUG
printf("parse_node: %s\n", current_node.name.c_str);
#endif
if (current == '[') {
current = parse_attr(stream, &current_node.attributes);
}
if (current != '{') {
// tag is selfclosing
open_node(&current_node, output);
return current;
}
current = strip(stream);
open_node(&current_node, output);
while(current != '}' && current != EOF) {
current = parse_node(current, stream, output);
}
close_node(&current_node, output);
done:
return strip(stream);
}
// long options
static struct option long_options[] = {
{"output", required_argument, 0, 'o'},
{"help", no_argument, 0, '?'},
{}
};
void usage(int argc, char ** args) {
printf("usage: %s [-o <filename>] file [file...]\n", args[0]);
printf("-I\tAdd Path to include paths\n");
printf("--output\n");
printf("-o\tThe output file to write the html to\n");
printf("\tWhen missing this option, stdout is used instead\n");
printf("-v\tEnable verbose output\n");
printf("--help Print this usage\n");
}
int main(int argc, char ** args) {
int i = 0;
FILE* output = NULL;
// parse arguments
while((i = getopt_long(argc, args, "o:I:v", long_options, NULL)) != -1) {
switch(i) {
case 'o':
output = fopen(optarg, "w");
if (output == NULL) {
fprintf(stderr, "could not create output file: %s (%s)\n", optarg, strerror(errno));
return -1;
}
break;
case 'I':
// add include dir
include_add_path(optarg);
break;
case 'v':
verbose = 1;
break;
case '?':
usage(argc, args);
return -1;
default:
// should not be reachable
abort();
}
}
if (output == NULL) {
output = stdout;
// disable output when printing to stdout
verbose = 0;
}
char * filename= NULL;
// parse all the files
for(i = optind;i < argc; i++) {
filename = args[i];
if (verbose) printf("starting conversion of %s\n", filename);
FILE * handle = fopen(filename, "r");
if (handle == NULL) {
fprintf(stderr, "could not open \"%s\": %s\n", filename, strerror(errno));
continue;
}
int current = strip(handle);
// parse the complete file
while((current = parse_node(current, handle, output)) != EOF);
fclose(handle);
}
if (output != stdout) fclose(output);
if (verbose) printf("done compiling\n");
return 0;
}
+80
View File
@@ -0,0 +1,80 @@
/*
* string.c
*
* Created on: 08.08.2017
* Author: julian
*/
#include "nhtml_string.h"
#include <stdlib.h> // needs: malloc, free
#include <memory.h> // needs: memcpy, memset
#include <string.h> // needs: strlen, strdup
int string_append(string_t *str, char c) {
if (str->c_str == NULL) {
// new string, need to emit EOS
str->len++;
}
char * new_ptr = realloc(str->c_str, str->len+1);
if (new_ptr == NULL) {
errno = ENOMEM;
return -1;
}
// append char
str->c_str = new_ptr;
str->c_str[str->len-1] = c;
str->c_str[str->len] = 0; // make sure
str->len++;
return 0;
}
int string_concat(string_t * dest, string_t *src) {
if (src->c_str == NULL) {
return 0;
}
if (dest->c_str == NULL) {
dest->len++;
}
char * new_ptr = realloc(dest->c_str, dest->len + src->len-1);
if (new_ptr == NULL) {
errno = ENOMEM;
return -1;
}
memcpy(dest->c_str + dest->len-1, src->c_str, src->len);
return 0;
}
void string_destroy(string_t *s) {
free(s->c_str);
memset(s, 0, sizeof(string_t));
}
string_t string_copy(string_t old) {
string_t tmp = {}; // initialize with 0
if (old.len == 0) {
return tmp;
}
tmp.c_str = malloc(old.len);
if (tmp.c_str == NULL) {
// The Application will have to handle a out of mem
// situation here
return tmp;
}
tmp.len = old.len;
memcpy(tmp.c_str, old.c_str, tmp.len);
return tmp;
}
string_t string_from_cstr(const char * str) {
string_t temp;
temp.len = strlen(str)+1;
temp.c_str = strdup(str);
return temp;
}