xml-parser.c

yum-metadata-parser 一个用C写的yum仓库解析库

    NULL,      /* resolveEntity */
    NULL,      /* getEntity */
    NULL,      /* entityDecl */
    NULL,      /* notationDecl */
    NULL,      /* attributeDecl */
    NULL,      /* elementDecl */
    NULL,      /* unparsedEntityDecl */
    NULL,      /* setDocumentLocator */
    NULL,      /* startDocument */
    NULL,      /* endDocument */
    (startElementSAXFunc) primary_sax_start_element, /* startElement */
    (endElementSAXFunc) primary_sax_end_element,     /* endElement */
    NULL,      /* reference */
    (charactersSAXFunc) sax_characters,      /* characters */
    NULL,      /* ignorableWhitespace */
    NULL,      /* processingInstruction */
    NULL,      /* comment */
    sax_warning,      /* warning */
    sax_error,      /* error */
    sax_error,      /* fatalError */
};

void
sax_context_init (SAXContext *sctx,
                  const char *md_type,
                  CountFn count_callback,
                  PackageFn package_callback,
                  gpointer user_data,
                  GError **err)
{
    sctx->md_type = md_type;
    sctx->error = err;
    sctx->count_fn = count_callback;
    sctx->package_fn = package_callback;
    sctx->user_data = user_data;
    sctx->current_package = NULL;
    sctx->want_text = FALSE;
    sctx->text_buffer = g_string_sized_new (PACKAGE_FIELD_SIZE);
}

void
yum_xml_parse_primary (const char *filename,
                       CountFn count_callback,
                       PackageFn package_callback,
                       gpointer user_data,
                       GError **err)
{
    PrimarySAXContext ctx;
    SAXContext *sctx = &ctx.sctx;
    int rc;

    ctx.state = PRIMARY_PARSER_TOPLEVEL;
    ctx.current_dep_list = NULL;
    ctx.current_file = NULL;

    sax_context_init(sctx, "primary.xml", count_callback, package_callback,
                     user_data, err);

    xmlSubstituteEntitiesDefault (1);
    rc = xmlSAXUserParseFile (&primary_sax_handler, &ctx, filename);

    if (sctx->current_package) {
        g_warning ("Incomplete package lost");
        package_free (sctx->current_package);
    }

    g_string_free (sctx->text_buffer, TRUE);
}

/*****************************************************************************/


static void
parse_package (const char **attrs, Package *p)
{
    int i;
    const char *attr;
    const char *value;

    for (i = 0; attrs && attrs[i]; i++) {
        attr = attrs[i];
        value = attrs[++i];

        if (!strcmp (attr, "pkgid"))
            p->pkgId = g_string_chunk_insert (p->chunk, value);
        if (!strcmp (attr, "name"))
            p->name = g_string_chunk_insert (p->chunk, value);
        else if (!strcmp (attr, "arch"))
            p->arch = g_string_chunk_insert (p->chunk, value);
    }
}

typedef enum {
    FILELIST_PARSER_TOPLEVEL = 0,
    FILELIST_PARSER_PACKAGE,
} FilelistSAXContextState;

typedef struct {
    SAXContext sctx;

    FilelistSAXContextState state;

    PackageFile *current_file;
} FilelistSAXContext;

static void
filelist_parser_toplevel_start (FilelistSAXContext *ctx,
                                const char *name,
                                const char **attrs)
{
    SAXContext *sctx = &ctx->sctx;

    if (!strcmp (name, "package")) {
        g_assert (sctx->current_package == NULL);

        ctx->state = FILELIST_PARSER_PACKAGE;

        sctx->current_package = package_new ();
        parse_package (attrs, sctx->current_package);
    }

    else if (sctx->count_fn && !strcmp (name, "filelists")) {
        int i;
        const char *attr;
        const char *value;

        for (i = 0; attrs && attrs[i]; i++) {
            attr = attrs[i];
            value = attrs[++i];

            if (!strcmp (attr, "packages")) {
                sctx->count_fn (string_to_guint32_with_default (value, 0),
                                sctx->user_data);
                break;
            }
        }
    }
}

static void
filelist_parser_package_start (FilelistSAXContext *ctx,
                               const char *name,
                               const char **attrs)
{
    SAXContext *sctx = &ctx->sctx;

    Package *p = sctx->current_package;
    int i;
    const char *attr;
    const char *value;

    g_assert (p != NULL);

    sctx->want_text = TRUE;

    if (!strcmp (name, "version")) {
        parse_version_info(attrs, p);
    }

    else if (!strcmp (name, "file")) {
        ctx->current_file = package_file_new ();

        for (i = 0; attrs && attrs[i]; i++) {
            attr = attrs[i];
            value = attrs[++i];

            if (!strcmp (attr, "type"))
                ctx->current_file->type =
                    g_string_chunk_insert_const (p->chunk, value);
        }
    }
}

static void
filelist_sax_start_element (void *data, const char *name, const char **attrs)
{
    FilelistSAXContext *ctx = (FilelistSAXContext *) data;
    SAXContext *sctx = &ctx->sctx;

    if (sctx->text_buffer->len)
        g_string_truncate (sctx->text_buffer, 0);

    switch (ctx->state) {
    case FILELIST_PARSER_TOPLEVEL:
        filelist_parser_toplevel_start (ctx, name, attrs);
        break;
    case FILELIST_PARSER_PACKAGE:
        filelist_parser_package_start (ctx, name, attrs);
        break;
    default:
        break;
    }
}

static void
filelist_parser_package_end (FilelistSAXContext *ctx, const char *name)
{
    SAXContext *sctx = &ctx->sctx;

    Package *p = sctx->current_package;

    g_assert (p != NULL);

    sctx->want_text = FALSE;

    if (!strcmp (name, "package")) {
        if (sctx->package_fn && !*sctx->error)
            sctx->package_fn (p, sctx->user_data);

        package_free (p);
        sctx->current_package = NULL;

        if (ctx->current_file) {
            g_free (ctx->current_file);
            ctx->current_file = NULL;
        }

        ctx->state = FILELIST_PARSER_TOPLEVEL;
    }

    else if (!strcmp (name, "file")) {
        PackageFile *file = ctx->current_file;
        file->name = g_string_chunk_insert_len (p->chunk,
                                                sctx->text_buffer->str,
                                                sctx->text_buffer->len);
        if (!file->type)
            file->type = g_string_chunk_insert_const (p->chunk, "file");

        p->files = g_slist_prepend (p->files, file);
        ctx->current_file = NULL;
    }
}

static void
filelist_sax_end_element (void *data, const char *name)
{
    FilelistSAXContext *ctx = (FilelistSAXContext *) data;
    SAXContext *sctx = &ctx->sctx;

    switch (ctx->state) {
    case FILELIST_PARSER_PACKAGE:
        filelist_parser_package_end (ctx, name);
        break;
    default:
        break;
    }

    g_string_truncate (sctx->text_buffer, 0);
}

static xmlSAXHandler filelist_sax_handler = {
    NULL,      /* internalSubset */
    NULL,      /* isStandalone */
    NULL,      /* hasInternalSubset */
    NULL,      /* hasExternalSubset */
    NULL,      /* resolveEntity */
    NULL,      /* getEntity */
    NULL,      /* entityDecl */
    NULL,      /* notationDecl */
    NULL,      /* attributeDecl */
    NULL,      /* elementDecl */
    NULL,      /* unparsedEntityDecl */
    NULL,      /* setDocumentLocator */
    NULL,      /* startDocument */
    NULL,      /* endDocument */
    (startElementSAXFunc) filelist_sax_start_element, /* startElement */
    (endElementSAXFunc) filelist_sax_end_element,     /* endElement */
    NULL,      /* reference */
    (charactersSAXFunc) sax_characters,      /* characters */
    NULL,      /* ignorableWhitespace */
    NULL,      /* processingInstruction */
    NULL,      /* comment */
    sax_warning,      /* warning */
    sax_error,      /* error */
    sax_error,      /* fatalError */
};

void
yum_xml_parse_filelists (const char *filename,
                         CountFn count_callback,
                         PackageFn package_callback,
                         gpointer user_data,
                         GError **err)
{
    FilelistSAXContext ctx;
    SAXContext *sctx = &ctx.sctx;

    int rc;

    ctx.state = FILELIST_PARSER_TOPLEVEL;
    ctx.current_file = NULL;
    
    sax_context_init(sctx, "filelists.xml", count_callback, package_callback,
                     user_data, err);

    xmlSubstituteEntitiesDefault (1);
    rc = xmlSAXUserParseFile (&filelist_sax_handler, &ctx, filename);

    if (sctx->current_package) {
        g_warning ("Incomplete package lost");
        package_free (sctx->current_package);
    }

    if (ctx.current_file)
        g_free (ctx.current_file);

    g_string_free (sctx->text_buffer, TRUE);
}

/*****************************************************************************/

typedef enum {
    OTHER_PARSER_TOPLEVEL = 0,
    OTHER_PARSER_PACKAGE,
} OtherSAXContextState;

typedef struct {
    SAXContext sctx;

    OtherSAXContextState state;

    ChangelogEntry *current_entry;
} OtherSAXContext;

static void
other_parser_toplevel_start (OtherSAXContext *ctx,
                             const char *name,
                             const char **attrs)
{
    SAXContext *sctx = &ctx->sctx;

    if (!strcmp (name, "package")) {
        g_assert (sctx->current_package == NULL);

        ctx->state = OTHER_PARSER_PACKAGE;

        sctx->current_package = package_new ();
        parse_package (attrs, sctx->current_package);
    }

    else if (sctx->count_fn && !strcmp (name, "otherdata")) {
        int i;
        const char *attr;
        const char *value;

        for (i = 0; attrs && attrs[i]; i++) {
            attr = attrs[i];
            value = attrs[++i];

            if (!strcmp (attr, "packages")) {
                sctx->count_fn (string_to_guint32_with_default (value, 0),
                                sctx->user_data);
                break;
            }
        }
    }
}

static void
other_parser_package_start (OtherSAXContext *ctx,
                            const char *name,
                            const char **attrs)
{
    SAXContext *sctx = &ctx->sctx;

    Package *p = sctx->current_package;
    int i;
    const char *attr;
    const char *value;

    g_assert (p != NULL);

    sctx->want_text = TRUE;

    if (!strcmp (name, "version")) {
        parse_version_info(attrs, p);
    }

    else if (!strcmp (name, "changelog")) {
        ctx->current_entry = changelog_entry_new ();

        for (i = 0; attrs && attrs[i]; i++) {
            attr = attrs[i];
            value = attrs[++i];

            if (!strcmp (attr, "author"))
                ctx->current_entry->author =
                    g_string_chunk_insert_const (p->chunk, value);
            else if (!strcmp (attr, "date"))
                ctx->current_entry->date = strtol(value, NULL, 10);
        }
    }
}

static void
other_sax_start_element (void *data, const char *name, const char **attrs)
{
    OtherSAXContext *ctx = (OtherSAXContext *) data;
    SAXContext *sctx = &ctx->sctx;

    if (sctx->text_buffer->len)
        g_string_truncate (sctx->text_buffer, 0);

    switch (ctx->state) {
    case OTHER_PARSER_TOPLEVEL:
        other_parser_toplevel_start (ctx, name, attrs);
        break;
    case OTHER_PARSER_PACKAGE:
        other_parser_package_start (ctx, name, attrs);
        break;
    default:
        break;
    }
}

static void
other_parser_package_end (OtherSAXContext *ctx, const char *name)
{
    SAXContext *sctx = &ctx->sctx;

    Package *p = sctx->current_package;

    g_assert (p != NULL);

    sctx->want_text = FALSE;

    if (!strcmp (name, "package")) {

        if (p->changelogs)
            p->changelogs = g_slist_reverse (p->changelogs);

        if (sctx->package_fn && !*sctx->error)
            sctx->package_fn (p, sctx->user_data);

        package_free (p);
        sctx->current_package = NULL;

        if (ctx->current_entry) {
            g_free (ctx->current_entry);
            ctx->current_entry = NULL;
        }

        ctx->state = OTHER_PARSER_TOPLEVEL;
    }

    else if (!strcmp (name, "changelog")) {
        ctx->current_entry->changelog =
            g_string_chunk_insert_len (p->chunk,
                                       sctx->text_buffer->str,
                                       sctx->text_buffer->len);

        p->changelogs = g_slist_prepend (p->changelogs, ctx->current_entry);
        ctx->current_entry = NULL;
    }
}

static void
other_sax_end_element (void *data, const char *name)
{
    OtherSAXContext *ctx = (OtherSAXContext *) data;
    SAXContext *sctx = &ctx->sctx;

    switch (ctx->state) {
    case OTHER_PARSER_PACKAGE:
        other_parser_package_end (ctx, name);
        break;
    default:
        break;
    }

    g_string_truncate (sctx->text_buffer, 0);
}

static xmlSAXHandler other_sax_handler = {
    NULL,      /* internalSubset */
    NULL,      /* isStandalone */
    NULL,      /* hasInternalSubset */
    NULL,      /* hasExternalSubset */
    NULL,      /* resolveEntity */
    NULL,      /* getEntity */
    NULL,      /* entityDecl */
    NULL,      /* notationDecl */
    NULL,      /* attributeDecl */
    NULL,      /* elementDecl */
    NULL,      /* unparsedEntityDecl */
    NULL,      /* setDocumentLocator */
    NULL,      /* startDocument */
    NULL,      /* endDocument */
    (startElementSAXFunc) other_sax_start_element, /* startElement */
    (endElementSAXFunc) other_sax_end_element,     /* endElement */
    NULL,      /* reference */
    (charactersSAXFunc) sax_characters,      /* characters */
    NULL,      /* ignorableWhitespace */
    NULL,      /* processingInstruction */
    NULL,      /* comment */
    sax_warning,      /* warning */
    sax_error,      /* error */
    sax_error,      /* fatalError */
};

void
yum_xml_parse_other (const char *filename,
                     CountFn count_callback,
                     PackageFn package_callback,
                     gpointer user_data,
                     GError **err)
{
    OtherSAXContext ctx;
    SAXContext *sctx = &ctx.sctx;

    int rc;

    ctx.state = OTHER_PARSER_TOPLEVEL;
    ctx.current_entry = NULL;
    
    sax_context_init(sctx, "other.xml", count_callback, package_callback,
                     user_data, err);

    xmlSubstituteEntitiesDefault (1);
    rc = xmlSAXUserParseFile (&other_sax_handler, &ctx, filename);

    if (sctx->current_package) {
        g_warning ("Incomplete package lost");
        package_free (sctx->current_package);
    }

    if (ctx.current_entry)
        g_free (ctx.current_entry);

    g_string_free (sctx->text_buffer, TRUE);
}