Annotation-System/hmmlconv/hmmlconv.c

#if 0
cc "$0" -g -std=c99 -D_POSIX_SOURCE -o "${0%.c}"
exit
#endif

#include <stdlib.h>
#include <stdio.h>
#include <errno.h>
#include <unistd.h>    // access
#include <string.h>    // strerror
#include <fcntl.h>     // open
#include <sys/types.h> // open, mkdir
#include <sys/stat.h>  // open, mkdir
#include <assert.h>

typedef struct
{
    char *Site;
    char *Page;
    char *URL;
} Resource;

static Resource Resources[64];
static int ResCount;

// Removes directories from a path, e.g.: /path/to/thing -> thing
char *BaseName(char *Path)
{
    char *Base = Path;
    for(char *c = Path; *c; ++c)
    {
        if(*c == '/')
        {
            Base = c + 1;
        }
    }

    return Base;
}

int PromptOverwrite(char *Name, int* Always)
{
    while(1){
        printf("The file '%s' exists. Overwrite? [Yes|No|Always|eXit]\n> ", Name);
        fflush(stdout);

        int C;
        switch((C = getchar()))
        {
            case 'a':
            case 'A':
                *Always = 1;
            case 'y':
            case 'Y':
                while(getchar() != '\n');
                return 1;
            case 'n':
            case 'N':
                while(getchar() != '\n');
                return 0;
            case 'x':
            case 'X':
            case EOF:
                exit(0);
        }

        printf("Unknown option '%c'\n", C);

        if(C != '\n'){
            while(getchar() != '\n');
        }
    }
}

char *ReadWholeFile(FILE *File)
{
    fseek(File, 0, SEEK_END);
    size_t Size = ftell(File);
    fseek(File, 0, SEEK_SET);
    
    char *Buffer = malloc(Size + 1);
    if(!Buffer)
    {
        perror("malloc");
        exit(1);
    }

    fread(Buffer, Size, 1, File);
    Buffer[Size] = 0;

    return Buffer;
}

void SkipWhitespace(char **Ptr)
{
    while(**Ptr && **Ptr <= ' ')
    {
        ++*Ptr;
    }
}

char *InPlaceUnescape(char *In)
{
    if(*In == '"') ++In;
    char *Result = In;
    char *Out = In;
    
    while(*In && *In != '"')
    {
        if(*In == '\\')
        {
            ++In;
        }

        *Out++ = *In++;
    }

    *Out = 0;
    return Result;
}

int IsAlNum(char C)
{
    return (C >= '0' && C <= '9') || (C >= 'a' && C <= 'z') || (C >= 'A' && C <= 'Z');
}

Resource *LookupResource(char *Tag, char *Line)
{
    Resource *Res = NULL;

    for(int i = 0; i < ResCount; ++i)
    {
        if(strcmp(Tag, Resources[i].Site) == 0 || strcmp(Tag, Resources[i].Page) == 0)
        {
            Res = Resources + i;
            if(strstr(Line, Res->Page))
            {
                break;
            }
        }
    }
    
    return Res;
}

void ProcessAnnotation(char *Line, FILE *OutFile)
{
    SkipWhitespace(&Line);
    assert(*Line == '"');

    fputc('[', OutFile);
    while(*++Line != '"')
    {
        fputc(*Line, OutFile);
    }
    fputc(']', OutFile);

    assert(*++Line == ':');
    assert(*++Line == ' ');
    assert(*++Line == '"');

    Line = InPlaceUnescape(Line);

    // convert author
    if(Line[0] == '@')
    {
        char *P = strchr(Line, ' ');
        assert(P);

        if(P[-1] == ':')
        {
            --P;
        }
        
        fprintf(OutFile, "[@%.*s]", (int)(P - Line), Line);
        Line = P+1;
    }

    char RefBuf[256];
    char *LinePtr;
    char *RunStart = Line;
    char *FirstSpace = NULL;
    int ConsiderQuote = 0;

    fputc('[', OutFile);
    
    int QuoteID = -1;
    
    for(LinePtr = Line; *LinePtr; ++LinePtr)
    {
        int ScanBytes = 0;
        int TmpQuoteID;
        
        // convert Resource -> ref
        if(LinePtr[0] == ' ' && LinePtr[1] == '[' && sscanf(LinePtr+1, "[see Resources, %255[^]]]%n", RefBuf, &ScanBytes) == 1 && ScanBytes)
        {
            //printf("Find Resource [%s]\n", RefBuf);
            Resource *Res = LookupResource(RefBuf, Line);
            if(Res){
                fprintf(OutFile,
                        "%.*s[ref\n    site=\"%s\"\n    page=\"%s\"\n    url=\"%s\"]",
                        (int)(LinePtr - RunStart),
                        RunStart,
                        Res->Site,
                        Res->Page,
                        Res->URL);

                LinePtr += ScanBytes;
                RunStart = LinePtr+1;

            }
            else
            {
                fprintf(stderr, "WARNING: can't find resource [%s] :(\n", RefBuf);
            }
        }

        // convert quotes
        else if(LinePtr[0] == ' ' && LinePtr[1] == '(' && sscanf(LinePtr+1, "(!quote %d)%n", &TmpQuoteID, &ScanBytes) == 1 && ScanBytes)
        {
            QuoteID = TmpQuoteID;
            fprintf(OutFile, "%.*s", (int)(LinePtr - RunStart), RunStart);
            LinePtr += ScanBytes;
            RunStart = LinePtr+1;
        }

        // Used for getting name for Q:'s
        else if(*LinePtr == ' ')
        {
            if(!FirstSpace)
            {
                FirstSpace = LinePtr;
                ConsiderQuote = 1;
            }
            else
            {    
                ConsiderQuote = 0;
            }
        }

        // Q:
        else if(LinePtr[0] == 'Q' && LinePtr[1] == ':' && ConsiderQuote)
        {
            LinePtr += 2;
            RunStart = LinePtr+1;
            fprintf(OutFile, "@%.*s][", (int)(FirstSpace - Line), Line);
        }
        
        // Escape stuff
        else if(*LinePtr == ']' ||
            (LinePtr[0] == '\\' && LinePtr[1] != '"') ||
            (LinePtr > Line && LinePtr[-1] == ' ' && strchr(":~@", LinePtr[0]) && !IsAlNum(LinePtr[1])))
        {
            fprintf(OutFile, "%.*s\\", (int)(LinePtr - RunStart), RunStart);
            RunStart = LinePtr;
        }
    }

    // write out remaining text
    fprintf(OutFile, "%.*s]", (int)(LinePtr - RunStart), RunStart);
    
    // write out quote node if applicable
    if(QuoteID != -1)
    {
        fprintf(OutFile, "[quote %d]", QuoteID);
    }
    
    fputc('\n', OutFile);
}

int ProcessFile(char *InFileName, FILE *InFile, FILE *OutFile)
{
    char *Contents = ReadWholeFile(InFile);
    char *Ptr;

    printf("Processing [%s]...\n", InFileName);

    // Resources
    ResCount = 0;
    if((Ptr = strstr(Contents, "\n## Resources")))
    {
        char *LinePtrState;
        char *LinePtr = strtok_r(Ptr+13, "\r\n", &LinePtrState);

        for(; LinePtr; LinePtr = strtok_r(NULL, "\r\n", &LinePtrState))
        {
            if(LinePtr[0] == '#' && LinePtr[1] == '#')
            {
                break;
            }

            if(LinePtr[0] != '*' || LinePtr[1] != ' ')
            {
                continue;
            }
            
            LinePtr += 2;

            // Site (seems to be optional [day 205])
            {
                int Inc = 2;
                char *Separator = strstr(LinePtr, " [");
                if(!Separator)
                {
                    Separator = strstr(LinePtr, " '[");
                    Inc = 3;
                }

                if(!Separator)
                {
                    Separator = strstr(LinePtr, "[");
                    Inc = 1;
                }

                if(!Separator) continue;

                char *S = (Separator[-1] == ',' || Separator[-1] == ':') ? Separator - 1 : Separator;
                *S = 0;

                Resources[ResCount].Site = LinePtr;
                LinePtr = Separator + Inc;
            }

            // Page
            {
                char *Separator = strstr(LinePtr, "](");
                if(!Separator) continue;

                // needed for day115, maybe others
                {
                    char *S = Separator[-1] == '*' ? Separator - 1 : Separator;
                    *S = 0;
                }
                if(*LinePtr == '*') ++LinePtr;

                Resources[ResCount].Page = LinePtr;
                LinePtr = Separator + 2;
            }

            // URL
            {
                char *Separator = strstr(LinePtr, ")");
                if(!Separator) continue;
                *Separator = 0;

                Resources[ResCount].URL = LinePtr;
                LinePtr = Separator + 1;
            }
            
            printf("Add Res: %s\n", Resources[ResCount].Site);

            ++ResCount;
        }
    }

    char *Title;
    char *VideoID;

    enum {
        STATE_METADATA,
        STATE_MARKERS,
    } State = STATE_METADATA;

    int LineNumber = 0;
    char *LineState;
    char *LinePtr = strtok_r(Contents, "\r\n", &LineState);

    for(; LinePtr; LinePtr = strtok_r(NULL, "\r\n", &LineState))
    {
        switch(State)
        {
            case STATE_METADATA: {
                if(strncmp(LinePtr, "title: ", 7) == 0)
                {
                    Title = InPlaceUnescape(LinePtr + 7);
                }
                else if(strncmp(LinePtr, "videoId: ", 9) == 0)
                {
                    VideoID = InPlaceUnescape(LinePtr + 9);
                }
                else if(strncmp(LinePtr, "markers:", 8) == 0)
                {
                    fprintf(OutFile, "[video member=cmuratori project=hero title=\"%s\" platform=youtube id=%s annotator=Miblo]\n", Title, VideoID);
                    State = STATE_MARKERS;
                }
            } break;

            case STATE_MARKERS: {
                if(strncmp(LinePtr, "---", 3) == 0){
                    goto Done;
                } else {
                    ProcessAnnotation(LinePtr, OutFile);
                }
            } break;
        }

        ++LineNumber;
    }

Done:
    fputs("[/video]\n", OutFile); 
    free(Contents);
}

int main(int ArgC, char **Args)
{
    if(ArgC < 3)
    {
        fprintf(stderr, "Usage: %s [Files...] [Output Directory]\n", Args[0]);
        return 1;
    }

    const char *OutDirName = Args[ArgC-1];

    // Check if the directory exists and we can write to it.
    // If it doesn't exist, create it.

    if(access(OutDirName, R_OK | W_OK | X_OK) == -1)
    {
        if(errno == ENOENT)
        {
            if(mkdir(OutDirName, 0777) == -1)
            {
                fprintf(stderr, "Couldn't create directory [%s]: %s\n", OutDirName, strerror(errno));
            }
        }
        else
        {
            fprintf(stderr, "Error accessing %s: %s\n", OutDirName, strerror(errno));
            return 1;
        }
    }

    // Loop through all the files and convert them.

    int Errors = 0;
    int AlwaysOverwrite = 0;

    for(int i = 1; i < ArgC-1; ++i)
    {
        char *FileNameBase = BaseName(Args[i]);
        char OutNameBuf[strlen(OutDirName) + strlen(FileNameBase) + 7];
        sprintf(OutNameBuf, "%s/%s.hmml", OutDirName, FileNameBase);

        FILE *OutFile;
        {
            int OpenFlags = O_CREAT | O_EXCL | O_WRONLY;

            // Use the POSIX open function for a change, and so we can more easily
            // test for the file already existing.
OpenOutput:;
            int OutFileDesc = open(OutNameBuf, OpenFlags, 0666);
            if(OutFileDesc == -1)
            {
                if(errno == EEXIST)
                {
                    if(AlwaysOverwrite || PromptOverwrite(OutNameBuf, &AlwaysOverwrite))
                    {
                        OpenFlags = (OpenFlags & ~O_CREAT) | O_TRUNC;
                        goto OpenOutput;
                    }
                }
                else
                {
                    perror("open");
                }
            } else {
                OutFile = fdopen(OutFileDesc, "w");
                if(!OutFile)
                {
                    fprintf(stderr, "Error opening %s: %s\n", OutNameBuf, strerror(errno));
                }
            }
        }

        FILE* InFile = fopen(Args[i], "r");
        if(!InFile)
        {
            fprintf(stderr, "Error opening %s: %s\n", Args[i], strerror(errno));
        }

        if(!InFile || !OutFile || !ProcessFile(FileNameBase, InFile, OutFile))
        {
            ++Errors;
        }

        if(InFile)  fclose(InFile);
        if(OutFile) fclose(OutFile);
    }

    if(Errors){
        printf("There were errors processing %d files.\n", Errors);
    }

    return Errors != 0;
}