struct-vis/main.c

//
// @TODO
// long-term goals:
//      - support C++!
//      - support Rust with repr(C) structs to the extent that it is possible!
//      - support Golang to the extent that it is possible!
//      - support Odin?!
//
// short-term:
//      - bitfields!
//      - robustness! change [64] byte name fields to pointers!
//      - find all files in folder of a given type!
//


#define STB_C_LEXER_DEFINITIONS

#define STB_C_LEX_0_IS_EOF             Y  // if Y, ends parsing at '\0'; if N, returns '\0' as token
#define STB_C_LEX_USE_STDLIB           Y  // use strtod,strtol for parsing #s; otherwise inaccurate hack
#define STB_C_LEX_DOLLAR_IDENTIFIER    N  // allow $ as an identifier character

#define STB_C_LEX_DEFINE_ALL_TOKEN_NAMES  Y   // if Y, all CLEX_ token names are defined, even if never returned
                                              // leaving it as N should help you catch config bugs

#define STB_C_LEX_DISCARD_PREPROCESSOR    Y   // discard C-preprocessor directives (e.g. after prepocess
                                              // still have #line, #pragma, etc)

#define STB_C_LEX_MULTILINE_DSTRINGS   N  // allow newlines in double-quoted strings
#define STB_C_LEX_MULTILINE_SSTRINGS   N  // allow newlines in single-quoted strings
#define STB_C_LEX_FLOAT_NO_DECIMAL     N  // allow floats that have no decimal point if they have an exponent


#define STB_C_LEX_C_IDENTIFIERS     Y   //  "[_a-zA-Z][_a-zA-Z0-9]*"               CLEX_id
#define STB_C_LEX_C_COMMENTS        Y   //  "/* comment */"
#define STB_C_LEX_CPP_COMMENTS      Y   //  "// comment to end of line\n"

#define STB_C_LEX_INTEGERS_AS_DOUBLES  N  // parses integers as doubles so they can be larger than 'int', but only if STB_C_LEX_STDLIB==N
#define STB_C_LEX_C_DECIMAL_INTS    N   //  "0|[1-9][0-9]*"                        CLEX_intlit
#define STB_C_LEX_C_HEX_INTS        N   //  "0x[0-9a-fA-F]+"                       CLEX_intlit
#define STB_C_LEX_C_OCTAL_INTS      N   //  "[0-7]+"                               CLEX_intlit
#define STB_C_LEX_C_DECIMAL_FLOATS  N   //  "[0-9]*(.[0-9]*([eE][-+]?[0-9]+)?)     CLEX_floatlit
#define STB_C_LEX_C99_HEX_FLOATS    N   //  "0x{hex}+(.{hex}*)?[pP][-+]?{hex}+     CLEX_floatlit
#define STB_C_LEX_C_DQ_STRINGS      N   //  double-quote-delimited strings with escapes  CLEX_dqstring
#define STB_C_LEX_C_SQ_STRINGS      N   //  single-quote-delimited strings with escapes  CLEX_ssstring
#define STB_C_LEX_C_CHARS           N   //  single-quote-delimited character with escape CLEX_charlits
#define STB_C_LEX_C_COMPARISONS     N   //  "==" CLEX_eq  "!=" CLEX_noteq   "<=" CLEX_lesseq  ">=" CLEX_greatereq
#define STB_C_LEX_C_LOGICAL         N   //  "&&"  CLEX_andand   "||"  CLEX_oror
#define STB_C_LEX_C_SHIFTS          N   //  "<<"  CLEX_shl      ">>"  CLEX_shr
#define STB_C_LEX_C_INCREMENTS      N   //  "++"  CLEX_plusplus "--"  CLEX_minusminus
#define STB_C_LEX_C_ARROW           N   //  "->"  CLEX_arrow
#define STB_C_LEX_EQUAL_ARROW       N   //  "=>"  CLEX_eqarrow
#define STB_C_LEX_C_BITWISEEQ       N   //  "&="  CLEX_andeq    "|="  CLEX_oreq     "^="  CLEX_xoreq
#define STB_C_LEX_C_ARITHEQ         N   //  "+="  CLEX_pluseq   "-="  CLEX_minuseq
                                        //  "*="  CLEX_muleq    "/="  CLEX_diveq    "%=" CLEX_modeq
                                        //  if both STB_C_LEX_SHIFTS & STB_C_LEX_ARITHEQ:
                                        //                      "<<=" CLEX_shleq    ">>=" CLEX_shreq

#define STB_C_LEX_PARSE_SUFFIXES    N   // letters after numbers are parsed as part of those numbers, and must be in suffix list below
#define STB_C_LEX_DECIMAL_SUFFIXES  ""  // decimal integer suffixes e.g. "uUlL" -- these are returned as-is in string storage
#define STB_C_LEX_HEX_SUFFIXES      ""  // e.g. "uUlL"
#define STB_C_LEX_OCTAL_SUFFIXES    ""  // e.g. "uUlL"
#define STB_C_LEX_FLOAT_SUFFIXES    ""  //

#define STB_C_LEXER_IMPLEMENTATION
#include "stb_c_lexer.h"

#include <inttypes.h> // strtoimax
#include <limits.h>
#include <assert.h> //assert
#include <stdio.h> // fread, fseek, ftell
#include <stdlib.h> // malloc, free
#include <stdarg.h> // va_start, va_list, va_end
#include <stdint.h> // intxx_t, etc.
#include <string.h> // memcmp
#include <stdbool.h> // bool


struct Dummy {
    char *p;
    char c;
    int x;
};

static inline void die(const char* format, ...) {
    va_list args;
    va_start(args, format);
    vprintf(format, args);
    va_end(args);
    exit(1);
}

static inline char* readWholeFile(const char* filepath, size_t *outSize) {
    FILE *fp = fopen(filepath, "rb");
    if (fp == NULL) {
        die("failed to open file: %s", filepath);
    }
    fseek(fp, 0, SEEK_END);
    size_t size = ftell(fp);
    fseek(fp, 0L, SEEK_SET);
    char *buffer = (char*) malloc(size + 1);
    fread(buffer, sizeof (char), size, fp);
    buffer[size] = '\0';
    fclose(fp);

    if (outSize != NULL) *outSize = size;

    return buffer;
}

static inline bool isWhitespace(char c) {
    return c == ' ' || c == '\r' || c == '\n' || c == '\f' || c == '\t';
}

static inline char* eatWhitespace(char* input) {
    char* orig = input;
    char c;
    while ((c = *input) != '\0') {
        if (!isWhitespace(c)) return input;
        input++;
    }
    return orig;
}

// de-duplicates whitespace
static inline char* findNthLastCharOccurence(char* string, int length, char c, int n) {
    char* out = NULL;
    int _n = 0;
    for (int i = length - 1; i > 0; i--) {
        if (string[i] == c) _n++;
        if (_n == n) return string + i;
        while (isWhitespace(string[i]) && i > 0) {
            i--;
        }
    }
    return out;
}

static inline int strWrite(char *dest, const char *src, int maxCount) {
    int i = 0;
    for (; i < maxCount; i++) {
        if (src[i] == '\0') {
            break;
        }

        dest[i] = src[i];
    }
    dest[i] = '\0';
    return i;
}

static ssize_t alignForward(ssize_t ptr, ssize_t align) {
    ssize_t p, a, modulo;
    p = ptr;
    a = align;
    modulo = p % a;
    if (modulo != 0) {
        p += a - modulo;
    }
    return p;
}

struct Declaration {
    char type[64];
    char name[64];
    ssize_t size;
    ssize_t align;
    bool isBitfield;
};

struct StructInfo {
    char name[64];
    char alias[64];
    const char *filename;
    int lineNumber, lineOffset;

    ssize_t size;
    struct Declaration declarations[16];
    int numDeclarations;
};

static inline void printStructInfo(struct StructInfo *structInfo) {
    printf("%s - %d:%d", structInfo->filename, structInfo->lineNumber, structInfo->lineOffset);
    printf(" - %s", structInfo->name[0] == '\0' ? "(anonymous struct)" : structInfo->name);
    printf(", %s\n", structInfo->alias[0] == '\0' ? "(c++ style, no typedef alias)" : structInfo->alias);
    printf(" - total size: %ld\n", structInfo->size);
    for (int i = 0; i < structInfo->numDeclarations; i++) {
        struct Declaration *decl = structInfo->declarations + i;
        printf("\tdecl name: %s, type: %s, size: %ld, alignment: %ld\n", decl->name, decl->type, decl->size, decl->align);
    }
}

static int capacityAllStructs = 64;
static int numAllStructs = 0;
static struct StructInfo *allStructs;

void pushStructInfo(struct StructInfo *structInfo) {
    if (numAllStructs >= capacityAllStructs) {
        capacityAllStructs *= 1.5;
        allStructs = realloc(allStructs, sizeof(struct StructInfo) * (capacityAllStructs));
    }

    memcpy((void*) &allStructs[numAllStructs++], (void*) structInfo, sizeof(struct StructInfo));
}

#include "table.h"
#include "visualization.h"


#define STORE_SIZE 1024*1000
static const int store_size = STORE_SIZE;
static char store[STORE_SIZE] = { 0 };
#undef STORE_SIZE

struct Array {
    unsigned int length;
    unsigned int capacity;
    void* data;
};
void push(struct Array* array, void* item) {

}

void parseType() {

}

static inline bool shouldSkipConst(char* nullTerminated) {
    // @HACK skip all instances of 'const'
    size_t bounds = sizeof("const");
    for (int i = 0; i < bounds; i++) {
        char c = nullTerminated[i];
        if (c != "const"[i]) return false;
    }

    return true;
}

static inline void finalizeDeclaration(
    char lineBuffer[128],
    int lookback,
    int numAsterisks,
    int numDeclarations,
    int arrayVal,
    struct StructInfo *structInfo
) {
    printf("LINE BUFFER: |%s|, arrayVal: %d\n", lineBuffer, arrayVal);
    // we're at the end of a line of declarations.
    // we can learn some interesting stuff by looking back now.
    char typeBuffer[64] = { 0 };
    char *cursor = findNthLastCharOccurence(lineBuffer, 128, ' ', lookback);
    if (cursor == NULL) {
        die("panic when finalizing a declaration");
    }

    struct Declaration *decl = structInfo->declarations + structInfo->numDeclarations;
    int diff = (int)(cursor - lineBuffer);
    int count = strWrite(typeBuffer, lineBuffer, diff);
    int multiplier = 1;
    if (arrayVal != -1) {
        multiplier = arrayVal;
    }

    ssize_t totalSize = 0;
    TableEntry *entry = lookup(typeTable, typeBuffer);
    if (numAsterisks == 0) {
        if (entry == NULL) {
            // this is likely a new/unknown type in the program. enter it into the type table with an unknown size.
            printf("warning: unknown field size and alignment in struct field: %s\n", typeBuffer);
            insertPadZeroes(typeTable, typeBuffer, -1, -1);
            decl->size = -1;
            decl->align = -1;

        } else {
            decl->size = entry->size * multiplier;
            decl->align = entry->align;
        }
    } else {
        decl->size = sizeof(void*) * multiplier;
        decl->align = sizeof(void*);
    }

    // we could have multiple declarations (comma separated)
    // they will have to be the same type, except for bitfields (kill me)
    // so we'll just copy the type from the first decl, and just move the cursor
    // to find the other name.
    for (int i = 0; i < numDeclarations; i++) {
        decl = structInfo->declarations + structInfo->numDeclarations;
        totalSize += decl->size;
        structInfo->numDeclarations++;

        // write in the type name field.
        // for looking up size in the table, we don't want to include the '*'
        // but for storing the type name of the decl, we probably do.
        for (int i = 0; i < numAsterisks; i++) {
            count += strWrite(typeBuffer + count, "*", 1);
        }
        strWrite(decl->type, typeBuffer, 64);

        // figure out the name of this field.
        char* nameStart;
        char c;
        while ((c = *cursor) != '\0') {
            if (!isWhitespace(c)) {
                nameStart = cursor;
                break;
            }
            cursor++;
        }
        char* nameEnd;
        while ((c = *cursor) != '\0') {
            if (isWhitespace(c)) {
                nameEnd = cursor;
                break;
            }
            cursor++;
        }
        int count = strWrite(decl->name, nameStart, (int) (nameEnd-nameStart));
        if (arrayVal != -1) {
            snprintf(decl->name + count, 64 - count, "[%d]", arrayVal);
        }
    }

    structInfo->size += totalSize;
}

void parseStructDeclaration(struct StructInfo *structInfo, stb_lexer *lexer) {
    bool somethingWasConst = false;
    bool numDeclarations = 1;
    int numAsterisks = 0;
    int soFar = 0;
    int lookback = 2;

    // for parsing things like 'char name[12]'
    char* lastOpenBracket = NULL;
    int arrayVal = -1;

    char lineBuffer[128] = { 0 };
    do {
        switch (lexer->token) {
            case 260: {
                // we don't record const because it's annoying.
                if (shouldSkipConst(lexer->string)) { somethingWasConst = true; break; }

                soFar += strWrite(lineBuffer + soFar, lexer->string, 64);
                soFar += strWrite(lineBuffer + soFar, " ", 1);
            } break;

            case ',':
                numDeclarations++;
                lookback++;
                break;

            case '*':
                numAsterisks++;
                break;

            case '[':
                lastOpenBracket = lexer->where_firstchar;
                break;

            case ']':
                arrayVal = strtoimax(lastOpenBracket + 1, &lexer->where_firstchar, 10);
                // happens if there is no value between, in which case it's a "name[]" decl
                if (arrayVal == 0) arrayVal = -1; 
                break;

            case ';': {
                finalizeDeclaration(lineBuffer, lookback, numAsterisks, numDeclarations, arrayVal, structInfo);
            } return;
        }
    } while (stb_c_lexer_get_token(lexer) != 0);
}

//
// the token in the lexer is a 'struct' keyword. we want to get the identifiers, and the nested declarations.
//
// <struct-or-union-specifier> ::= <struct-or-union> <identifier> { {<struct-declaration>}+ }
//                               | <struct-or-union> { {<struct-declaration>}+ }
//                               | <struct-or-union> <identifier>
void parseStruct(const char *filename, stb_lexer *lexer, bool isClass) {
    int result = stb_c_lexer_get_token(lexer);
    if (result == 0) die("failed to parse struct");

    stb_lex_location location = { 0 };
    stb_c_lexer_get_location(lexer, lexer->where_firstchar, &location);

    struct StructInfo structInfo = { 0 };
    structInfo.filename        = filename;
    structInfo.lineNumber      = location.line_number;
    structInfo.lineOffset      = location.line_offset;
    structInfo.numDeclarations = 0;
    structInfo.size            = 0;

    switch (lexer->token) {
        case 260: {
            char tempNameBuffer[64] = { 0 };
            strWrite(tempNameBuffer, lexer->string, 64);

            // maybe a named struct.
            result = stb_c_lexer_get_token(lexer);
            if (result == 0) die("failed to parse struct");

            if (lexer->token == '{') {
                strWrite(structInfo.name, tempNameBuffer, 64);

            } else {
                return;
            }
        } break;

        case '{': {} break;

        default: return;
    }

    int balancer = 1;
    while (stb_c_lexer_get_token(lexer) != 0) {
        switch (lexer->token) {
            case '}': if (--balancer == 0) goto checkTypeAlias;
            case '{':     ++balancer;      break;

            case 260: {
                parseStructDeclaration(&structInfo, lexer);
            } break;
        }
    }

checkTypeAlias:
    result = stb_c_lexer_get_token(lexer);
    if (result == 0) die("unexpected end of stream when parsing a struct");

    if (lexer->token == 260) {
        // we have a type alias for the struct.
        // @NOTE @TODO this could also conceivably by the __attribute__ thingy: https://stackoverflow.com/questions/14671253/is-there-a-gcc-keyword-to-allow-structure-reordering
        strWrite(structInfo.alias, lexer->string, 64);
    }

    pushStructInfo(&structInfo);
}

void parseTypedef(stb_lexer *lexer) {

}


void parseFile(const char *filepath) {
    printf("parsing file %s...\n", filepath);
    size_t size;
    char *buffer = readWholeFile(filepath, &size);

    stb_lexer lexer;
    stb_c_lexer_init(&lexer, buffer, buffer + size + 1, store, store_size);

    while (stb_c_lexer_get_token(&lexer) != 0) {
        switch (lexer.token) {
            case 260: { // token is a string
                const uint64_t LE_STRUCT  = 0x0000746375727473U;
                const uint64_t LE_CLASS   = 0x0000007373616C63U;
                const uint64_t LE_TYPEDEF = 0x0066656465707974U;

                uint64_t t = *((uint64_t*)(lexer.string));
                     if ((t                     ) == LE_TYPEDEF) { parseTypedef(&lexer); }
                else if ((t & 0x00FFFFFFFFFFFFFF) == LE_STRUCT)  { parseStruct(filepath, &lexer, false); }
                else if ((t & 0x0000FFFFFFFFFFFF) == LE_CLASS)   { parseStruct(filepath, &lexer, true); }
            } break;
        }
    }
    free(buffer);
    memset(store, 0, store_size);
}

// http://www.catb.org/esr/structure-packing/
int main(int argc, char* argv[]) {
    // @TODO check for flag -fshort-enums
    allStructs = malloc(sizeof(struct StructInfo) * capacityAllStructs);
    typeTable = initTable();

    if (CHAR_BIT != 8) {
        printf("warning - CHAR_BIT != 8\n");
    }

    if (false) {
        printf("CHAR_BIT       = %d\n", CHAR_BIT);
        printf("MB_LEN_MAX     = %d\n\n", MB_LEN_MAX);

        printf("CHAR_MIN       = %+d\n", CHAR_MIN);
        printf("CHAR_MAX       = %+d\n", CHAR_MAX);
        printf("SCHAR_MIN      = %+d\n", SCHAR_MIN);
        printf("SCHAR_MAX      = %+d\n", SCHAR_MAX);
        printf("UCHAR_MAX      = %u\n\n", UCHAR_MAX);

        printf("SHRT_MIN       = %+d\n", SHRT_MIN);
        printf("SHRT_MAX       = %+d\n", SHRT_MAX);
        printf("USHRT_MAX      = %u\n\n", USHRT_MAX);

        printf("INT_MIN        = %+d\n", INT_MIN);
        printf("INT_MAX        = %+d\n", INT_MAX);
        printf("UINT_MAX       = %u\n\n", UINT_MAX);

        printf("LONG_MIN       = %+ld\n", LONG_MIN);
        printf("LONG_MAX       = %+ld\n", LONG_MAX);
        printf("ULONG_MAX      = %lu\n\n", ULONG_MAX);

        printf("LLONG_MIN      = %+lld\n", LLONG_MIN);
        printf("LLONG_MAX      = %+lld\n", LLONG_MAX);
        printf("ULLONG_MAX     = %llu\n\n", ULLONG_MAX);

        printf("PTRDIFF_MIN    = %td\n", PTRDIFF_MIN);
        printf("PTRDIFF_MAX    = %+td\n", PTRDIFF_MAX);
        printf("SIZE_MAX       = %zu\n", SIZE_MAX);
        printf("SIG_ATOMIC_MIN = %+jd\n",(intmax_t)SIG_ATOMIC_MIN);
        printf("SIG_ATOMIC_MAX = %+jd\n",(intmax_t)SIG_ATOMIC_MAX);
        printf("WCHAR_MIN      = %+jd\n",(intmax_t)WCHAR_MIN);
        printf("WCHAR_MAX      = %+jd\n",(intmax_t)WCHAR_MAX);
        printf("WINT_MIN       = %jd\n", (intmax_t)WINT_MIN);
        printf("WINT_MAX       = %jd\n", (intmax_t)WINT_MAX);
    }

    if (argc < 2) {
        //die("provide a list of c/c++ files and/or headers to anaylze.");
        parseFile(__FILE__);
        parseFile("table.h");
        parseFile("visualization.h");
        parseFile("stb_c_lexer.h");
    }

    for (int i = 1; i < argc; i++) {
        const char *filepath = argv[i];
        parseFile(filepath);
    }

    outputHtml();

    return 0;
}