1

     2

#include <stdlib.h>

     3

     4

#include "lex.h"

     5

#include "error.h"

     6

#include "log.h"

     7

     8

#define INITIAL_BUF_SIZE 4096

     9

    10

int lexer (const struct lex *lex, const char *input, void *arg) {

    11

    // handling error returns

    12

    int err = -1, cb_err;

    13

    14

    // token buffer

    15

    char *buf = NULL, *buf_ptr;

    16

    size_t buf_size = INITIAL_BUF_SIZE;

    17

    18

    // state

    19

    int prev_state = LEX_INITIAL, cur_state = lex->initial_state, next_state = LEX_INITIAL;

    20

    21

    // input chars

    22

    const char *c = input;

    23

    24

    // lookups

    25

    const struct lex_transition *trans = NULL;

    26

    27

    // allocate the buffer

    28

    if ((buf = malloc(sizeof(char) * buf_size)) == NULL)

    29

        goto error;

    30

    31

    // set buf_ptr initial position

    32

    buf_ptr = buf;

    33

    34

    // clear input

    35

    DEBUG("*cough*");

    36

    DEBUGN("%s", "");

    37

    38

    // process input

    39

    do {

    40

        if (*c) {

    41

            // look up the next state

    42

            for (trans = lex->state_list[cur_state - 1].trans_list; trans->next_state > 0 || trans->flags; trans++) {

    43

                // accept defaults

    44

                if (trans->flags & LEX_TRANS_DEFAULT)

    45

                    break;

    46

    47

                // disregard non-matches

    48

                if (trans->left > *c || *c > trans->right)

    49

                    continue;

    50

    51

                // abort on invalids

    52

                if (trans->flags & LEX_TRANS_INVALID) {

    53

                    goto error;

    54

    55

                } else {

    56

                    // accept it

    57

                    break;

    58

                }

    59

            }

    60

    61

            // did we find a transition with a valid next state?

    62

            if (!(next_state = trans->next_state))

    63

                goto error;

    64

    65

            // call the char handler

    66

            if (lex->char_fn && (cb_err = lex->char_fn(*c, cur_state, next_state, arg)))

    67

                goto error;

    68

    69

        } else {

    70

            // EOF!

    71

            next_state = LEX_EOF;

    72

    73

            // is cur_state a valid end state?

    74

            if (!(lex->state_list[cur_state - 1].flags & LEX_STATE_END))

    75

                goto error;

    76

    77

            // note: we don't pass the NUL byte to the char handler

    78

        }

    79

    80

        // if this char is part of the next token...

    81

        if (next_state != cur_state) {

    82

            // terminate the buffer and reset buf_ptr

    83

            *buf_ptr = 0; buf_ptr = buf;

    84

    85

            // dump state transitions

    86

            DEBUGF("\n\t%25s -> %25s -> %25s",

    87

                LEX_STATE_NAME(lex, prev_state),

    88

                LEX_STATE_NAME(lex, cur_state),

    89

                LEX_STATE_NAME(lex, next_state)

    90

            );

    91

    92

            // pass in the complete token to the handler

    93

            if (lex->token_fn && (cb_err = lex->token_fn(cur_state, buf, next_state, prev_state, arg)))

    94

                goto error;

    95

    96

            // update states

    97

            prev_state = cur_state;

    98

            cur_state = next_state;

    99

            next_state = LEX_INITIAL;

   100

        }

   101

   102

        // dump chars

   103

        if (next_state == LEX_INITIAL)

   104

            DEBUGN("%c", *c);

   105

        else

   106

            DEBUGNF("%c", *c);

   107

   108

        // store this char in the buffer

   109

        *(buf_ptr++) = *c;

   110

   111

        // grow the buffer if needed

   112

        if (buf_ptr - buf >= buf_size) {

   113

            // remember the offset, as buf_ptr might get invalidated if buf is moved

   114

            size_t buf_offset = buf_ptr - buf;

   115

   116

            // calc new size

   117

            buf_size *= 2;

   118

   119

            // grow/move

   120

            if ((buf = realloc(buf, buf_size)) == NULL)

   121

                goto error;

   122

   123

            // fix buf_ptr

   124

            buf_ptr = buf + buf_offset;

   125

        }

   126

    } while (*(c++));

   127

   128

    // call the end handler

   129

    if (lex->end_fn && (cb_err = lex->end_fn(cur_state, arg)))

   130

        goto error;

   131

   132

    // successfully parsed!

   133

    err = 0;

   134

   135

error:

   136

    DEBUGNF("\n");

   137

   138

    if (cb_err)

   139

        err = cb_err;

   140

   141

    // dump debug info on error

   142

    if (err) {

   143

        const char *cc;

   144

   145

        // figure out the error

   146

        if (!buf)

   147

            WARNING("malloc/realloc");

   148

   149

        else if (trans && trans->flags & LEX_TRANS_INVALID)

   150

            WARNING("hit invalid transition match");

   151

   152

        else if (!next_state)

   153

            WARNING("no valid transition found");

   154

   155

        else if (next_state == LEX_EOF && !(lex->state_list[cur_state - 1].flags & LEX_STATE_END))

   156

            WARNING("invalid end state");

   157

   158

        else

   159

            WARNING("unknown error condition (!?)");

   160

   161

        DEBUG("%s", input);

   162

        DEBUGN("%s", "");

   163

   164

        for (cc = input; cc < c; cc++)

   165

            DEBUGNF(" ");

   166

   167

        DEBUGF("^\t%s -> %s -> %s",

   168

            LEX_STATE_NAME(lex, prev_state),

   169

            LEX_STATE_NAME(lex, cur_state),

   170

            LEX_STATE_NAME(lex, next_state)

   171

        );

   172

    }

   173

   174

    // free stuff

   175

    free(buf);

   176

   177

    // return

   178

    return err;

   179

}

   180

   181