terom@16: 
terom@16: #include <stdlib.h>
terom@15: 
terom@15: #include "lex.h"
terom@16: #include "error.h"
terom@16: #include "log.h"
terom@16: 
terom@16: #define INITIAL_BUF_SIZE 4096
terom@15: 
terom@15: int lexer (const struct lex *lex, const char *input, void *arg) {
terom@16:     // handling error returns
terom@16:     int err = -1, cb_err;
terom@16:     
terom@16:     // token buffer
terom@16:     char *buf = NULL, *buf_ptr;
terom@16:     size_t buf_size = INITIAL_BUF_SIZE;
terom@16:     
terom@16:     // state
terom@16:     int prev_state = LEX_INITIAL, cur_state = lex->initial_state, next_state = LEX_INITIAL;
terom@16:     
terom@16:     // input chars
terom@16:     const char *c = input;
terom@16: 
terom@16:     // lookups
terom@16:     const struct lex_transition *trans = NULL;
terom@16: 
terom@16:     // allocate the buffer
terom@16:     if ((buf = malloc(sizeof(char) * buf_size)) == NULL)
terom@16:         goto error;
terom@16: 
terom@16:     // set buf_ptr initial position
terom@16:     buf_ptr = buf;
terom@16:     
terom@16:     // clear input
terom@16:     DEBUG("*cough*");
terom@16:     DEBUGN("%s", "");
terom@16: 
terom@16:     // process input
terom@16:     do {
terom@16:         if (*c) {
terom@16:             // look up the next state
terom@16:             for (trans = lex->state_list[cur_state - 1].trans_list; trans->next_state > 0; trans++) {
terom@16:                 // accept defaults
terom@16:                 if (trans->flags & LEX_TRANS_DEFAULT)
terom@16:                     break;
terom@16:                 
terom@16:                 // disregard non-matches
terom@16:                 if (trans->left > *c || *c > trans->right)
terom@16:                     continue;
terom@16:                 
terom@16:                 // abort on invalids
terom@16:                 if (trans->flags & LEX_TRANS_INVALID)
terom@16:                     goto error;
terom@16:                 
terom@16:                 else {
terom@16:                     // accept it
terom@16:                     break;
terom@16:                 }
terom@16:             }
terom@16:             
terom@16:             // did we find a transition with a valid next state?
terom@16:             if (!(next_state = trans->next_state))
terom@16:                 goto error;
terom@16: 
terom@16:             // call the char handler
terom@16:             if (lex->char_fn && (cb_err = lex->char_fn(*c, cur_state, next_state, arg)))
terom@16:                 goto error;
terom@16: 
terom@16:         } else {
terom@16:             // EOF!
terom@16:             next_state = LEX_EOF;
terom@16:             
terom@16:             // is cur_state a valid end state?
terom@16:             if (!(lex->state_list[cur_state - 1].flags & LEX_STATE_END))
terom@16:                 goto error;
terom@16:             
terom@16:             // note: we don't pass the NUL byte to the char handler
terom@16:         }
terom@16: 
terom@16:         // if this char is part of the next token...
terom@16:         if (next_state != cur_state) {
terom@16:             // terminate the buffer and reset buf_ptr
terom@16:             *buf_ptr = 0; buf_ptr = buf;
terom@16:             
terom@16:             // dump state transitions
terom@16:             DEBUGF("\n\t%25s -> %25s -> %25s",
terom@16:                 LEX_STATE_NAME(lex, prev_state),
terom@16:                 LEX_STATE_NAME(lex, cur_state),
terom@16:                 LEX_STATE_NAME(lex, next_state)
terom@16:             );
terom@16: 
terom@16:             // pass in the complete token to the handler
terom@16:             if (lex->token_fn && (cb_err = lex->token_fn(cur_state, buf, next_state, prev_state, arg)))
terom@16:                 goto error;
terom@16: 
terom@16:             // update states
terom@16:             prev_state = cur_state;
terom@16:             cur_state = next_state;
terom@16:             next_state = LEX_INITIAL;
terom@16:         }
terom@16:         
terom@16:         // dump chars
terom@16:         if (next_state == LEX_INITIAL)
terom@16:             DEBUGN("%c", *c);
terom@16:         else
terom@16:             DEBUGNF("%c", *c);
terom@16:         
terom@16:         // store this char in the buffer
terom@16:         *(buf_ptr++) = *c;
terom@16: 
terom@16:         // grow the buffer if needed
terom@16:         if (buf_ptr - buf >= buf_size) {
terom@16:             // remember the offset, as buf_ptr might get invalidated if buf is moved
terom@16:             size_t buf_offset = buf_ptr - buf;
terom@16: 
terom@16:             // calc new size
terom@16:             buf_size *= 2;
terom@16:             
terom@16:             // grow/move
terom@16:             if ((buf = realloc(buf, buf_size)) == NULL)
terom@16:                 goto error;
terom@16:             
terom@16:             // fix buf_ptr
terom@16:             buf_ptr = buf + buf_offset;
terom@16:         }
terom@16:     } while (*(c++));
terom@16: 
terom@16:     // call the end handler
terom@16:     if (lex->end_fn && (cb_err = lex->end_fn(cur_state, arg)))
terom@16:         goto error;
terom@16: 
terom@16:     // successfully parsed!
terom@16:     err = 0;
terom@16: 
terom@16: error:
terom@16:     DEBUGNF("\n");
terom@16:     
terom@16:     if (cb_err)
terom@16:         err = cb_err;
terom@16: 
terom@16:     // dump debug info on error
terom@16:     if (err) {
terom@16:         const char *cc;
terom@16:         
terom@16:         // figure out the error
terom@16:         if (!buf)
terom@16:             WARNING("malloc/realloc");
terom@16: 
terom@16:         else if (trans && trans->flags & LEX_TRANS_INVALID)
terom@16:             WARNING("hit invalid transition match");
terom@16: 
terom@16:         else if (!next_state)
terom@16:             WARNING("no valid transition found");
terom@16:             
terom@16:         else if (next_state == LEX_EOF && !(lex->state_list[cur_state - 1].flags & LEX_STATE_END))
terom@16:             WARNING("invalid end state");
terom@16:         
terom@16:         else
terom@16:             WARNING("unknown error condition (!?)");
terom@16: 
terom@16:         DEBUG("%s", input);
terom@16:         DEBUGN("%s", "");
terom@16: 
terom@16:         for (cc = input; cc < c; cc++)
terom@16:             DEBUGNF(" ");
terom@16: 
terom@16:         DEBUGF("^\t%s -> %s -> %s",
terom@16:             LEX_STATE_NAME(lex, prev_state),
terom@16:             LEX_STATE_NAME(lex, cur_state),
terom@16:             LEX_STATE_NAME(lex, next_state)
terom@16:         );
terom@16:     }
terom@16: 
terom@16:     // free stuff
terom@16:     free(buf);
terom@16: 
terom@16:     // return
terom@16:     return err;
terom@15: }
terom@15: 
terom@16: