finit ?

Makefile

@@ -1,5 +1,5 @@
 BIN=webpage
-SRCS=main.c config.c page.c lexer.c
+SRCS=main.c config.c page.c lexer.c eval.c
 INC=.
 LIB=curl
 FLAGS=-Wall -Wextra -Og -g -ggdb -fvar-tracking
@@ -12,7 +12,11 @@ $(BIN) : $(SRCS)
 run : $(BIN)
 	./$(BIN) $(input)
 
+check :
+	cppcheck --enable=all --suppress=missingIncludeSystem -I$(INC) .
+	flawfinder .
+
 clean :
 	rm -f $(BIN) *.o
 
-.PHONY : run clean
+.PHONY : run check clean

eval.c

@@ -0,0 +1,142 @@
+#include "eval.h"
+#include "format.h"
+#include "lexer.h"
+#include "form.h"
+
+// TODO: BLOCKQUOTE OL PROGRESS CODE TABLE
+
+#define PAGE_WIDTH 80
+
+#define EXPAND_LIT(x) x, sizeof(x)
+
+PageState state = {0};
+
+void print_text(const char* text, int len){
+  for (int i=0; i<len; i++){
+    if (state.beginLine) printf("\n");
+
+    if (text[i] != '\n'){
+      printf("%c", text[i]);
+      state.y++;
+      state.beginLine = 0;
+    }
+
+    if (!state.beginLine && (state.y >= PAGE_WIDTH || text[i] == '\n')){
+      state.y = 0;
+      state.x++;
+      state.beginLine = 1;
+    }
+  }
+
+  state.beginLine = state.beginLine || text[len] == '\n';
+}
+
+void break_line(void){
+  state.y = 0;
+  state.x++;
+  state.beginLine = 1;
+}
+
+void evaluate(Token* token){
+  if (token->type == TITLE)
+  {
+    printf(FC_BOLD);
+    print_text(EXPAND_LIT("--- "));
+  }
+  else if (token->type == END_TITLE)
+  {
+    print_text(EXPAND_LIT(" ---"));
+    printf(FC_nBOLD);
+    break_line();
+  }
+  else if (token->type == UL){
+    state.inList = 1;
+  }
+  else if (token->type == UL){
+    state.inList = 0;
+  }
+  else if (token->type == LI)
+  {
+    break_line();
+    print_text(EXPAND_LIT("- "));
+  }
+  else if (token->type == END_LI)
+  {
+    //print_text(EXPAND_LIT("\n"));
+  }
+  else if (token->type == STRONG || token->type == B)
+  {
+    printf(FC_BOLD);
+  }
+  else if (token->type == END_STRONG || token->type == END_B)
+  {
+    printf(FC_nBOLD);
+    print_text(EXPAND_LIT(" "));
+  }
+  else if (token->type == EM || token->type == I)
+  {
+    printf(FC_ITA);
+  }
+  else if (token->type == END_EM || token->type == END_I)
+  {
+    printf(FC_nITA);
+  }
+  else if (token->type == IMG)
+  {
+    print_text(EXPAND_LIT("["));
+    print_text(EXPAND_LIT("IMG"));
+    print_text(EXPAND_LIT("]"));
+  }
+  else if (token->type == A)
+  {
+    print_text(EXPAND_LIT("["));
+    char* s = malloc(sizeof(int)*2);
+    sprintf(s, "%d", state.nbLink);
+    print_text(s, strlen(s));
+    state.nbLink++;
+    free(s);
+    print_text(EXPAND_LIT("]"));
+    printf(FC_UDL);
+  }
+  else if (token->type == END_A)
+  {
+    printf(FC_nUDL);
+    print_text(EXPAND_LIT(" "));
+  }
+  else if (token->type == P)
+  {
+    if (!state.inList) break_line();
+  }
+  else if (token->type == END_P)
+  {
+    break_line();
+  }
+  else if (token->type == TEXT)
+  {
+    print_text(token->value, token->len);
+  }
+  else if (token->type == H1 || token->type == H2 || token->type == H3 || token->type == H4 || token->type == H5 || token->type == H6)
+  {
+    break_line();
+    printf(FC_BOLD);
+  }
+  else if (token->type == END_H1 || token->type == END_H2 || token->type == END_H3 || token->type == END_H4 || token->type == END_H5 || token->type == END_H6)
+  {
+    printf(FC_nBOLD);
+  }
+  else if (token->type == HR)
+  {
+    printf(FC_nUDL);
+    break_line();
+    print_text(EXPAND_LIT("-------------------------------------------------------------------------------"));
+  }
+  else if (token->type == BLOCKQUOTE)
+  {
+    print_text(EXPAND_LIT("    "));
+    printf(FC_SWAP);
+  }
+  else if (token->type == END_BLOCKQUOTE)
+  {
+    printf(FC_nSWAP);
+  }
+}

eval.h

@@ -0,0 +1,16 @@
+#ifndef EVAL_H
+#define EVAL_H
+
+#include "lexer.h"
+
+typedef struct PageState {
+  int nbLink;
+  int x;
+  int y;
+  int beginLine;
+  int inList;
+} PageState;
+
+void evaluate(Token* token);
+
+#endif // EVAL_H

format.h

@@ -0,0 +1,60 @@
+#ifndef FORMAT_H
+#define FORMAT_H
+
+// COLOR
+
+#define FC_RES "\033[0m"
+#define FC_BOLD "\033[1m"
+#define FC_ITA "\033[3m"
+#define FC_UDL "\033[4m"
+#define FC_BIK "\033[5m"
+#define FC_SWAP "\033[7m"
+#define FC_HID "\033[8m"
+#define FC_STR "\033[9m"
+#define FC_DEF "\033[39m"
+#define FC_nBOLD "\033[22m"
+#define FC_nITA "\033[23m"
+#define FC_nUDL "\033[24m"
+#define FC_nBIK "\033[25m"
+#define FC_nSWAP "\033[27m"
+#define FC_nHID "\033[28m"
+#define FC_nSTR "\033[29m"
+#define FC_bDEF "\033[49m"
+
+#define FC_BLK "\033[30m"
+#define FC_RED "\033[31m"
+#define FC_GRN "\033[32m"
+#define FC_YEL "\033[33m"
+#define FC_BLU "\033[34m"
+#define FC_MAG "\033[35m"
+#define FC_CYA "\033[36m"
+#define FC_WHI "\033[37m"
+
+#define FC_lBLK "\033[90m"
+#define FC_lRED "\033[91m"
+#define FC_lGRN "\033[92m"
+#define FC_lYEL "\033[93m"
+#define FC_lBLU "\033[94m"
+#define FC_lMAG "\033[95m"
+#define FC_lCYA "\033[96m"
+#define FC_lWHI "\033[97m"
+
+#define FC_bBLK "\033[40m"
+#define FC_bRED "\033[41m"
+#define FC_bGRN "\033[42m"
+#define FC_bYEL "\033[43m"
+#define FC_bBLU "\033[44m"
+#define FC_bMAG "\033[45m"
+#define FC_bCYA "\033[46m"
+#define FC_bWHI "\033[47m"
+
+#define FC_blBLK "\033[100m"
+#define FC_blRED "\033[101m"
+#define FC_blGRN "\033[102m"
+#define FC_blYEL "\033[103m"
+#define FC_blBLU "\033[104m"
+#define FC_blMAG "\033[105m"
+#define FC_blCYA "\033[106m"
+#define FC_blWHI "\033[107m"
+
+#endif // FORMAT_H

lexer.c

@@ -1,6 +1,7 @@
 #include "lexer.h"
 
 #define HTML_BALISE_LEN 12
+#define DA_LEN 64
 
 typedef struct Cursor {
   int chunk;
@@ -17,68 +18,15 @@ Cursor prev = {
   .offset = -2,
 };
 
+TokenType token_by_name(const char name[HTML_BALISE_LEN]);
+
 void printtoken(Token* token){
   if (token == NULL){
     puts("NULL TOKEN");
     return;
   }
 
-  switch (token->type) {
-    case UNDEFINED_TYPE:
-      printf("UNDEFINED_TYPE: ");
-      break;
-    case DONT_CARE:
-      printf("DONT_CARE: ");
-      break;
-    case TEXT:
-      printf("TEXT: ");
-      break;
-    case BODY:
-      printf("BODY: ");
-      break;
-    case END_BODY:
-      printf("END_BODY: ");
-      break;
-    case HTML:
-      printf("HTML: ");
-      break;
-    case END_HTML:
-      printf("END_HTML: ");
-      break;
-    case A:
-      printf("A: ");
-      break;
-    case END_A:
-      printf("END_A: ");
-      break;
-    case UL:
-      printf("UL: ");
-      break;
-    case LI:
-      printf("LI: ");
-      break;
-    case H1:
-      printf("H1: ");
-      break;
-    case H2:
-      printf("H2: ");
-      break;
-    case H3:
-      printf("H3: ");
-      break;
-    case H4:
-      printf("H4: ");
-      break;
-    case H5:
-      printf("H5: ");
-      break;
-    case H6:
-      printf("H6: ");
-      break;
-    default:
-      printf("ERROR: UNKNOWN TOKEN: ");
-      break;
-  }
+  printf("%d: ", token->type);
 
   if (token->value == NULL){
     puts("'NO VALUE FOUND'");
@@ -126,60 +74,77 @@ void go_back(void){
   prev.offset--;
 
   if (curr.chunk < 0 || curr.offset < 0){
-    puts("ERROR: go way too back.");
+    puts("ERROR: cursor got way too back.");
     exit(1);
   }
 }
 
-TokenType token_by_name(const char name[HTML_BALISE_LEN]){
-  if (name == NULL){
-    return UNDEFINED_TYPE;
-  } else if (strncmp(name, "body", HTML_BALISE_LEN) == 0){
-    return BODY;
-  } else if (strncmp(name, "/body", HTML_BALISE_LEN) == 0){
-    return END_BODY;
-  } else if (strncmp(name, "html", HTML_BALISE_LEN) == 0){
-    return HTML;
-  } else if (strncmp(name, "/html", HTML_BALISE_LEN) == 0){
-    return END_HTML;
-  } else if (strncmp(name, "a", HTML_BALISE_LEN) == 0){
-    return A;
-  } else if (strncmp(name, "/a", HTML_BALISE_LEN) == 0){
-    return END_A;
-  } else if (strncmp(name, "ul", HTML_BALISE_LEN) == 0){
-    return UL;
-  } else if (strncmp(name, "li", HTML_BALISE_LEN) == 0){
-    return LI;
-  } else if (strncmp(name, "h1", HTML_BALISE_LEN) == 0){
-    return H1;
-  } else if (strncmp(name, "h2", HTML_BALISE_LEN) == 0){
-    return H2;
-  } else if (strncmp(name, "h3", HTML_BALISE_LEN) == 0){
-    return H3;
-  } else if (strncmp(name, "h4", HTML_BALISE_LEN) == 0){
-    return H4;
-  } else if (strncmp(name, "h5", HTML_BALISE_LEN) == 0){
-    return H5;
-  } else if (strncmp(name, "h6", HTML_BALISE_LEN) == 0){
-    return H6;
-  }
-
-  return DONT_CARE;
-}
-
-Token* create_text_token(Token* token, char* cursor){
-  int i = 0;
+char* getParam(const char* word, int len, char* cursor, int* size){
+  char* res = NULL;
+  int found = 0;
+  len--;
 
   do {
     cursor = nextchar();
+    for (int i=0; i<len; i++){
+      if (word[i] != *cursor){
+        found = -1;
+        break;
+      } else {
+        found = 1;
+        cursor = nextchar();
+      }
+    }
+
+    if (found == 1){
+      while (*cursor != '"'){
+        cursor = nextchar();
+      }
+
+      int cap = DA_LEN;
+      res = malloc(sizeof(char) * cap);
+      *size = 0;
+
+      do {
+        cursor = nextchar();
+        res[*size] = *cursor;
+        (*size)++;
+        if (*size >= cap){
+          cap *= 2;
+          res = realloc(res, cap);
+        }
+      } while (*cursor != '"');
+
+      res[*size-1] = '\0';
+
+      break;
+    }
+  } while (*cursor != '>');
+
+
+  return res;
+}
+
+Token* create_text_token(Token* token, char* cursor){
+  token = malloc(sizeof(Token));
+  token->value = malloc(sizeof(char) * DA_LEN);
+  int i = 0, cap = DA_LEN;
+
+  go_back();
+  do {
+    cursor = nextchar();
+    token->value[i] = *cursor;
     i++;
+    if (i >= cap){
+      cap *= 2;
+      token->value = realloc(token->value, cap);
+    }
   } while (*cursor != '<');
+  token->value[i-1] = '\0';
   go_back();
 
-  token = malloc(sizeof(Token));
   token->type = TEXT;
-  token->value = "TODO";
-  token->len = i;
+  token->len = i-1;
 
   return token;
 }
@@ -197,9 +162,27 @@ Token* create_balise_token(Token* token, char* cursor){
 
   token = malloc(sizeof(Token));
   token->type = token_by_name(balise);
-  token->value = malloc(sizeof(char) * len);
-  strncpy(token->value, balise, len+1);
-  token->len = len;
+
+  if (token->type == A){
+    token->value = getParam("href", sizeof("href"), cursor, &token->len);
+  } else if (token->type == IMG) {
+    int srclen, altlen, totallen;
+    char* src = getParam("src", sizeof("src"), cursor, &srclen);
+    char* alt = getParam("alt", sizeof("alt"), cursor, &altlen);
+    totallen = srclen * altlen + 1;
+
+    token->value = malloc(sizeof(char) * (totallen));
+    strncpy(token->value, src, srclen);
+    token->value[srclen-1] = ' ';
+    strncpy(token->value+srclen, alt, altlen);
+    token->value[totallen] = '\0';
+    token->len = totallen;
+
+  } else {
+    token->value = malloc(sizeof(char) * len);
+    strncpy(token->value, balise, len+1);
+    token->len = len;
+  }
 
   go_back();
   do {
@@ -228,3 +211,121 @@ Token* nexttoken(void){
 
   return token;
 }
+
+TokenType token_by_name(const char name[HTML_BALISE_LEN]){
+  if (name == NULL){
+    return UNDEFINED_TYPE;
+  } else if (strncmp(name, "body", HTML_BALISE_LEN) == 0){
+    return BODY;
+  } else if (strncmp(name, "/body", HTML_BALISE_LEN) == 0){
+    return END_BODY;
+  } else if (strncmp(name, "/html", HTML_BALISE_LEN) == 0){
+    return END_HTML;
+  } else if (strncmp(name, "title", HTML_BALISE_LEN) == 0){
+    return TITLE;
+  } else if (strncmp(name, "/title", HTML_BALISE_LEN) == 0){
+    return END_TITLE;
+  } else if (strncmp(name, "em", HTML_BALISE_LEN) == 0){
+    return EM;
+  } else if (strncmp(name, "/em", HTML_BALISE_LEN) == 0){
+    return END_EM;
+  } else if (strncmp(name, "i", HTML_BALISE_LEN) == 0){
+    return I;
+  } else if (strncmp(name, "/i", HTML_BALISE_LEN) == 0){
+    return END_I;
+  } else if (strncmp(name, "b", HTML_BALISE_LEN) == 0){
+    return B;
+  } else if (strncmp(name, "/b", HTML_BALISE_LEN) == 0){
+    return END_B;
+  } else if (strncmp(name, "strong", HTML_BALISE_LEN) == 0){
+    return STRONG;
+  } else if (strncmp(name, "/strong", HTML_BALISE_LEN) == 0){
+    return END_STRONG;
+  } else if (strncmp(name, "hr", HTML_BALISE_LEN) == 0){
+    return HR;
+  } else if (strncmp(name, "br", HTML_BALISE_LEN) == 0){
+    return BR;
+  } else if (strncmp(name, "p", HTML_BALISE_LEN) == 0){
+    return P;
+  } else if (strncmp(name, "/p", HTML_BALISE_LEN) == 0){
+    return END_P;
+  } else if (strncmp(name, "a", HTML_BALISE_LEN) == 0){
+    return A;
+  } else if (strncmp(name, "/a", HTML_BALISE_LEN) == 0){
+    return END_A;
+  } else if (strncmp(name, "ol", HTML_BALISE_LEN) == 0){
+    return OL;
+  } else if (strncmp(name, "/ol", HTML_BALISE_LEN) == 0){
+    return END_OL;
+  } else if (strncmp(name, "ul", HTML_BALISE_LEN) == 0){
+    return UL;
+  } else if (strncmp(name, "/ul", HTML_BALISE_LEN) == 0){
+    return END_UL;
+  } else if (strncmp(name, "li", HTML_BALISE_LEN) == 0){
+    return LI;
+  } else if (strncmp(name, "/li", HTML_BALISE_LEN) == 0){
+    return END_LI;
+  } else if (strncmp(name, "img", HTML_BALISE_LEN) == 0){
+    return IMG;
+  } else if (strncmp(name, "/img", HTML_BALISE_LEN) == 0){
+    return END_IMG;
+  } else if (strncmp(name, "blockquote", HTML_BALISE_LEN) == 0){
+    return BLOCKQUOTE;
+  } else if (strncmp(name, "/blockquote", HTML_BALISE_LEN) == 0){
+    return END_BLOCKQUOTE;
+  } else if (strncmp(name, "code", HTML_BALISE_LEN) == 0){
+    return CODE;
+  } else if (strncmp(name, "/code", HTML_BALISE_LEN) == 0){
+    return END_CODE;
+  } else if (strncmp(name, "h1", HTML_BALISE_LEN) == 0){
+    return H1;
+  } else if (strncmp(name, "h2", HTML_BALISE_LEN) == 0){
+    return H2;
+  } else if (strncmp(name, "h3", HTML_BALISE_LEN) == 0){
+    return H3;
+  } else if (strncmp(name, "h4", HTML_BALISE_LEN) == 0){
+    return H4;
+  } else if (strncmp(name, "h5", HTML_BALISE_LEN) == 0){
+    return H5;
+  } else if (strncmp(name, "h6", HTML_BALISE_LEN) == 0){
+    return H6;
+  } else if (strncmp(name, "/h1", HTML_BALISE_LEN) == 0){
+    return END_H1;
+  } else if (strncmp(name, "/h2", HTML_BALISE_LEN) == 0){
+    return END_H2;
+  } else if (strncmp(name, "/h3", HTML_BALISE_LEN) == 0){
+    return END_H3;
+  } else if (strncmp(name, "/h4", HTML_BALISE_LEN) == 0){
+    return END_H4;
+  } else if (strncmp(name, "/h5", HTML_BALISE_LEN) == 0){
+    return END_H5;
+  } else if (strncmp(name, "/h6", HTML_BALISE_LEN) == 0){
+    return H6;
+  } else if (strncmp(name, "table", HTML_BALISE_LEN) == 0){
+    return TABLE;
+  } else if (strncmp(name, "/table", HTML_BALISE_LEN) == 0){
+    return END_TABLE;
+  } else if (strncmp(name, "thead", HTML_BALISE_LEN) == 0){
+    return THEAD;
+  } else if (strncmp(name, "/thead", HTML_BALISE_LEN) == 0){
+    return END_THEAD;
+  } else if (strncmp(name, "tbody", HTML_BALISE_LEN) == 0){
+    return TBODY;
+  } else if (strncmp(name, "/tbody", HTML_BALISE_LEN) == 0){
+    return END_TBODY;
+  } else if (strncmp(name, "tr", HTML_BALISE_LEN) == 0){
+    return TR;
+  } else if (strncmp(name, "/tr", HTML_BALISE_LEN) == 0){
+    return END_TR;
+  } else if (strncmp(name, "th", HTML_BALISE_LEN) == 0){
+    return TH;
+  } else if (strncmp(name, "/th", HTML_BALISE_LEN) == 0){
+    return END_TH;
+  } else if (strncmp(name, "progress", HTML_BALISE_LEN) == 0){
+    return PROGRESS;
+  } else if (strncmp(name, "/progress", HTML_BALISE_LEN) == 0){
+    return END_PROGRESS;
+  }
+
+  return DONT_CARE;
+}

lexer.h

@@ -7,15 +7,27 @@
 
 #include "page.h"
 
+// 46 enum
 typedef enum TokenType {
   UNDEFINED_TYPE,
   DONT_CARE,
-  TEXT,
+  TITLE, END_TITLE,
+  TEXT, END_HTML,
   BODY, END_BODY,
-  HTML, END_HTML,
   A, END_A,
-  UL, LI,
+  IMG, END_IMG,
+  BLOCKQUOTE, END_BLOCKQUOTE,
+  CODE, END_CODE,
+  HR, BR, P, END_P,
+  PROGRESS, END_PROGRESS,
+  STRONG, B, I, EM,
+  END_STRONG, END_B, END_I, END_EM,
+  UL, OL, LI, END_UL, END_OL, END_LI,
+  TABLE, END_TABLE, THEAD,
+  END_THEAD, TBODY, END_TBODY,
+  TR, TH, END_TR, END_TH,
   H1, H2, H3, H4, H5, H6,
+  END_H1, END_H2, END_H3, END_H4, END_H5, END_H6,
 } TokenType;
 
 typedef struct Token {

main.c

@@ -1,18 +1,22 @@
 #include "config.h"
 #include "page.h"
 #include "lexer.h"
+#include "eval.h"
 
 int main(int argc, char* argv[]){
 
   getUserConfig(argc, argv);
   getPage();
 
+  //printPage();
+
   Token* token = NULL;
   do {
     token = nexttoken();
-    printtoken(token);
-    //evaluate(token);
-  } while (token != NULL && token->type != END_BODY);
+    evaluate(token);
+  } while (token != NULL && token->type != END_HTML);
+
+  printf("\n");
 
   return 0;
 }