Changes in uspace/lib/sif/src/sif.c [ee8d4d6:ca95ccd] in mainline

File:

• uspace/lib/sif/src/sif.c (modified) (39 diffs)

uspace/lib/sif/src/sif.c

@@ -1 +1 @@
 /*
- * Copyright (c) 2018 Jiri Svoboda
+ * Copyright (c) 2024 Jiri Svoboda
  * All rights reserved.
  *
@@ -33 +33 @@
  *
  * Structured Information Format (SIF) is an API that allows an application
- * to maintain data in a persistent repository in a format that
- *
- *  - is structured (and hence extensible)
- *  - allows atomic (transactional) updates
- *  - allows efficient updates
+ * to maintain data in a persistent repository in a format that is
+ * structured (and hence extensible).
  *
  * SIF is meant to be used as the basis for the storage backend used to
@@ -44 +41 @@
  * similar to an XML document that contains just tags with attributes
  * (but no text).
- *
- * SIF can thus naturally express ordered lists (unlike a relational database).
- * When contrasted to a (relational) database, SIF is much more primitive.
- *
- * In particular, SIF
- *
- *  - does not run on a separate server
- *  - des not handle concurrency
- *  - does not have a notion of types or data validation
- *  - does not understand relations
- *  - does not implement any kind of search/queries
- *  - does not deal with data sets large enough not to fit in primary memory
- *
- * any kind of structure data validation is left up to the application.
  */
 
 #include <adt/list.h>
 #include <adt/odict.h>
+#include <ctype.h>
 #include <errno.h>
+#include <stdbool.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -70 +55 @@
 
 static errno_t sif_export_node(sif_node_t *, FILE *);
-static errno_t sif_import_node(sif_node_t *, FILE *, sif_node_t **);
+static errno_t sif_import_node(sif_node_t *, FILE *, sif_node_t **, bool *);
 static sif_attr_t *sif_node_first_attr(sif_node_t *);
 static sif_attr_t *sif_node_next_attr(sif_attr_t *);
@@ -77 +62 @@
 static int sif_attr_cmp(void *, void *);
 
+/** Determine if character can start a name.
+ *
+ * @param c Character
+ * @return @c true iff the character can start a name
+ */
+static bool sif_is_name_start_char(char c)
+{
+        return isalpha(c) || c == '_';
+}
+
+/** Determine if character can continue a name.
+ *
+ * @param c Character
+ * @return @c true iff the character can continue a name
+ */
+static bool sif_is_name_char(char c)
+{
+        return isalnum(c) || c == '-' || c == '.';
+}
+
 /** Create new SIF node.
  *
@@ -172 +177 @@
 }
 
-/** Create and open a SIF repository.
- *
- * @param fname File name
- * @param rsess Place to store pointer to new session.
+/** Create SIF document.
+ *
+ * @param rdoc Place to store pointer to new document.
  *
  * @return EOK on success or error code
  */
-errno_t sif_create(const char *fname, sif_sess_t **rsess)
-{
-        sif_sess_t *sess;
+errno_t sif_new(sif_doc_t **rdoc)
+{
+        sif_doc_t *doc;
         sif_node_t *root = NULL;
-        sif_trans_t *trans = NULL;
         errno_t rc;
-        FILE *f;
-
-        sess = calloc(1, sizeof(sif_sess_t));
-        if (sess == NULL)
-                return ENOMEM;
-
-        sess->fname = str_dup(fname);
-        if (sess->fname == NULL) {
-                rc = ENOMEM;
-                goto error;
-        }
+
+        doc = calloc(1, sizeof(sif_doc_t));
+        if (doc == NULL)
+                return ENOMEM;
 
         root = sif_node_new(NULL);
@@ -209 +205 @@
         }
 
-        f = fopen(fname, "wx");
-        if (f == NULL) {
-                rc = EIO;
-                goto error;
-        }
-
-        sess->f = f;
-        sess->root = root;
-
-        /* Run a dummy trasaction to marshall initial repo state to file */
-        rc = sif_trans_begin(sess, &trans);
-        if (rc != EOK)
-                goto error;
-
-        rc = sif_trans_end(trans);
-        if (rc != EOK)
-                goto error;
-
-        *rsess = sess;
+        doc->root = root;
+
+        *rdoc = doc;
         return EOK;
 error:
-        if (trans != NULL)
-                sif_trans_abort(trans);
         sif_node_delete(root);
-        if (sess->fname != NULL)
-                free(sess->fname);
-        free(sess);
+        free(doc);
         return rc;
 }
 
-/** Open an existing SIF repository.
+/** Load SIF document.
  *
  * @param fname File name
- * @param rsess Place to store pointer to new session.
+ * @param rdoc Place to store pointer to new document.
  *
  * @return EOK on success or error code
  */
-errno_t sif_open(const char *fname, sif_sess_t **rsess)
-{
-        sif_sess_t *sess;
+errno_t sif_load(const char *fname, sif_doc_t **rdoc)
+{
+        sif_doc_t *doc;
         sif_node_t *root = NULL;
         errno_t rc;
-        FILE *f;
-
-        sess = calloc(1, sizeof(sif_sess_t));
-        if (sess == NULL)
-                return ENOMEM;
-
-        sess->fname = str_dup(fname);
-        if (sess->fname == NULL) {
+        bool endtag;
+        FILE *f = NULL;
+
+        doc = calloc(1, sizeof(sif_doc_t));
+        if (doc == NULL)
+                return ENOMEM;
+
+        doc->fname = str_dup(fname);
+        if (doc->fname == NULL) {
                 rc = ENOMEM;
                 goto error;
@@ -269 +246 @@
         }
 
-        rc = sif_import_node(NULL, f, &root);
-        if (rc != EOK)
+        rc = sif_import_node(NULL, f, &root, &endtag);
+        if (rc != EOK || endtag == true)
                 goto error;
 
@@ -278 +255 @@
         }
 
-        sess->root = root;
-
-        sess->f = f;
-        sess->root = root;
-        *rsess = sess;
+        fclose(f);
+        doc->root = root;
+        *rdoc = doc;
         return EOK;
 error:
         sif_node_delete(root);
-        if (sess->fname != NULL)
-                free(sess->fname);
-        free(sess);
+        free(doc);
+        if (f != NULL)
+                fclose(f);
         return rc;
 }
 
-/** Close SIF session.
- *
- * @param sess SIF session
+/** Delete SIF document.
+ *
+ * @param doc SIF document
  * @return EOK on success or error code
  */
-errno_t sif_close(sif_sess_t *sess)
-{
-        sif_node_delete(sess->root);
-
-        if (fclose(sess->f) < 0) {
-                free(sess);
-                return EIO;
-        }
-
-        if (sess->fname != NULL)
-                free(sess->fname);
-        free(sess);
-        return EOK;
+void sif_delete(sif_doc_t *doc)
+{
+        sif_node_delete(doc->root);
+        free(doc);
 }
 
 /** Return root node.
  *
- * @param sess SIF session
- */
-sif_node_t *sif_get_root(sif_sess_t *sess)
-{
-        return sess->root;
+ * @param doc SIF document
+ */
+sif_node_t *sif_get_root(sif_doc_t *doc)
+{
+        return doc->root;
 }
 
@@ -383 +349 @@
 }
 
-/** Begin SIF transaction.
- *
- * @param trans Transaction
+/** Save SIF document to file.
+ * *
+ * @param doc SIF document
+ * @param fname File name
  * @return EOK on success or error code
  */
-errno_t sif_trans_begin(sif_sess_t *sess, sif_trans_t **rtrans)
-{
-        sif_trans_t *trans;
-
-        trans = calloc(1, sizeof(sif_trans_t));
-        if (trans == NULL)
-                return ENOMEM;
-
-        trans->sess = sess;
-        *rtrans = trans;
-        return EOK;
-}
-
-/** Commit SIF transaction.
- *
- * Commit and free the transaction. If an error is returned, that means
- * the transaction has not been freed (and sif_trans_abort() must be used).
- *
- * @param trans Transaction
- * @return EOK on success or error code
- */
-errno_t sif_trans_end(sif_trans_t *trans)
-{
+errno_t sif_save(sif_doc_t *doc, const char *fname)
+{
+        FILE *f = NULL;
         errno_t rc;
 
-        (void) fclose(trans->sess->f);
-
-        trans->sess->f = fopen(trans->sess->fname, "w");
-        if (trans->sess->f == NULL)
-                return EIO;
-
-        rc = sif_export_node(trans->sess->root, trans->sess->f);
+        f = fopen(fname, "w");
+        if (f == NULL) {
+                rc = EIO;
+                goto error;
+        }
+
+        rc = sif_export_node(doc->root, f);
         if (rc != EOK)
-                return rc;
-
-        if (fputc('\n', trans->sess->f) == EOF)
-                return EIO;
-
-        if (fflush(trans->sess->f) == EOF)
-                return EIO;
-
-        free(trans);
-        return EOK;
-}
-
-/** Abort SIF transaction.
- *
- * @param trans Transaction
- */
-void sif_trans_abort(sif_trans_t *trans)
-{
-        free(trans);
+                goto error;
+
+        if (fflush(f) == EOF) {
+                rc = EIO;
+                goto error;
+        }
+
+        fclose(f);
+        return EOK;
+error:
+        if (f != NULL)
+                fclose(f);
+        return rc;
 }
 
@@ -447 +388 @@
  * @a parent.
  *
- * @param trans Transaction
  * @param parent Parent node
  * @param ctype Child type
@@ -454 +394 @@
  * @return EOK on success or ENOMEM if out of memory
  */
-errno_t sif_node_prepend_child(sif_trans_t *trans, sif_node_t *parent,
-    const char *ctype, sif_node_t **rchild)
+errno_t sif_node_prepend_child(sif_node_t *parent, const char *ctype,
+    sif_node_t **rchild)
 {
         sif_node_t *child;
@@ -479 +419 @@
  * Create a new child and append it at the end of children list of @a parent.
  *
- * @param trans Transaction
  * @param parent Parent node
  * @param ctype Child type
@@ -486 +425 @@
  * @return EOK on success or ENOMEM if out of memory
  */
-errno_t sif_node_append_child(sif_trans_t *trans, sif_node_t *parent,
-    const char *ctype, sif_node_t **rchild)
+errno_t sif_node_append_child(sif_node_t *parent, const char *ctype,
+    sif_node_t **rchild)
 {
         sif_node_t *child;
@@ -511 +450 @@
  * Create a new child and insert it before an existing child.
  *
- * @param trans Transaction
  * @param sibling Sibling before which to insert
  * @param ctype Child type
@@ -518 +456 @@
  * @return EOK on success or ENOMEM if out of memory
  */
-errno_t sif_node_insert_before(sif_trans_t *trans, sif_node_t *sibling,
-    const char *ctype, sif_node_t **rchild)
+errno_t sif_node_insert_before(sif_node_t *sibling, const char *ctype,
+    sif_node_t **rchild)
 {
         sif_node_t *child;
@@ -543 +481 @@
  * Create a new child and insert it after an existing child.
  *
- * @param trans Transaction
  * @param sibling Sibling after which to insert
  * @param ctype Child type
@@ -550 +487 @@
  * @return EOK on success or ENOMEM if out of memory
  */
-errno_t sif_node_insert_after(sif_trans_t *trans, sif_node_t *sibling,
-    const char *ctype, sif_node_t **rchild)
+errno_t sif_node_insert_after(sif_node_t *sibling, const char *ctype,
+    sif_node_t **rchild)
 {
         sif_node_t *child;
@@ -573 +510 @@
 /** Destroy SIF node.
  *
- * This function does not return an error, but the transaction may still
- * fail to complete.
- *
- * @param trans Transaction
  * @param node Node to destroy
  */
-void sif_node_destroy(sif_trans_t *trans, sif_node_t *node)
+void sif_node_destroy(sif_node_t *node)
 {
         list_remove(&node->lparent);
@@ -587 +520 @@
 /** Set node attribute.
  *
- * @param trans Transaction
  * @param node SIF node
  * @param aname Attribute name
@@ -594 +526 @@
  * @return EOK on success, ENOMEM if out of memory
  */
-errno_t sif_node_set_attr(sif_trans_t *trans, sif_node_t *node,
-    const char *aname, const char *avalue)
+errno_t sif_node_set_attr(sif_node_t *node, const char *aname,
+    const char *avalue)
 {
         odlink_t *link;
@@ -636 +568 @@
 /** Unset node attribute.
  *
- * This function does not return an error, but the transaction may still
- * fail to complete.
- *
- * @param trans Transaction
  * @param node Node
  * @param aname Attribute name
  */
-void sif_node_unset_attr(sif_trans_t *trans, sif_node_t *node,
-    const char *aname)
+void sif_node_unset_attr(sif_node_t *node, const char *aname)
 {
         odlink_t *link;
@@ -658 +585 @@
 }
 
-/** Export string to file.
- *
- * Export string to file (the string is bracketed and escaped).
+/** Export node name to file.
+ *
+ * Export node name to file.
  *
  * @param str String
@@ -666 +593 @@
  * @return EOK on success, EIO on I/O error
  */
+static errno_t sif_export_name(const char *str, FILE *f)
+{
+        if (fputs(str, f) == EOF)
+                return EIO;
+
+        return EOK;
+}
+
+/** Export string to file.
+ *
+ * Export string to file (the string is double-quoted and escaped).
+ *
+ * @param str String
+ * @param f File
+ * @return EOK on success, EIO on I/O error
+ */
 static errno_t sif_export_string(const char *str, FILE *f)
 {
         const char *cp;
 
-        if (fputc('[', f) == EOF)
+        if (fputc('"', f) == EOF)
                 return EIO;
 
         cp = str;
         while (*cp != '\0') {
-                if (*cp == ']' || *cp == '\\') {
-                        if (fputc('\\', f) == EOF)
+                if (*cp == '<') {
+                        if (fputs("&lt;", f) == EOF)
+                                return EIO;
+                } else if (*cp == '"') {
+                        if (fputs("&quot;", f) == EOF)
+                                return EIO;
+                } else {
+                        if (fputc(*cp, f) == EOF)
                                 return EIO;
                 }
-                if (fputc(*cp, f) == EOF)
+
+                ++cp;
+        }
+
+        if (fputc('"', f) == EOF)
+                return EIO;
+
+        return EOK;
+}
+
+/** Read characters from file, make sure they match the specified sequence.
+ *
+ * @param f File
+ * @param chars Expected sequence of characters to be read
+ *
+ * @return EOK on success, EIO on I/O error or mismatch
+ */
+static errno_t sif_get_verify_chars(FILE *f, const char *chars)
+{
+        const char *cp;
+        char c;
+
+        cp = chars;
+        while (*cp != '\0') {
+                c = fgetc(f);
+                if (c != *cp)
                         return EIO;
                 ++cp;
         }
 
-        if (fputc(']', f) == EOF)
-                return EIO;
-
-        return EOK;
-}
-
-/** Import string from file.
- *
- * Import string from file (the string in the file must be
- * properly bracketed and escaped).
- *
+        return EOK;
+}
+
+/** Import name from file.
+ * *
  * @param f File
  * @param rstr Place to store pointer to newly allocated string
  * @return EOK on success, EIO on I/O error
  */
-static errno_t sif_import_string(FILE *f, char **rstr)
+static errno_t sif_import_name(FILE *f, char **rstr)
 {
         char *str;
+        char *nstr;
         size_t str_size;
         size_t sidx;
@@ -714 +683 @@
 
         c = fgetc(f);
-        if (c != '[') {
+        if (!sif_is_name_start_char(c)) {
+                rc = EIO;
+                goto error;
+        }
+
+        while (true) {
+                if (sidx >= str_size) {
+                        str_size *= 2;
+                        nstr = realloc(str, str_size + 1);
+                        if (nstr == NULL) {
+                                rc = ENOMEM;
+                                goto error;
+                        }
+
+                        str = nstr;
+                }
+
+                str[sidx++] = c;
+
+                c = fgetc(f);
+                if (!sif_is_name_char(c))
+                        break;
+        }
+
+        ungetc(c, f);
+
+        str[sidx] = '\0';
+        *rstr = str;
+        return EOK;
+error:
+        free(str);
+        return rc;
+}
+
+/** Import string from file.
+ *
+ * Import string from file (the string in the file must be
+ * properly quoted and escaped).
+ *
+ * @param f File
+ * @param rstr Place to store pointer to newly allocated string
+ * @return EOK on success, EIO on I/O error
+ */
+static errno_t sif_import_string(FILE *f, char **rstr)
+{
+        char *str;
+        char *nstr;
+        size_t str_size;
+        size_t sidx;
+        int c;
+        errno_t rc;
+
+        str_size = 1;
+        sidx = 0;
+        str = malloc(str_size + 1);
+        if (str == NULL)
+                return ENOMEM;
+
+        c = fgetc(f);
+        if (c != '"') {
                 rc = EIO;
                 goto error;
@@ -726 +754 @@
                 }
 
-                if (c == ']')
+                if (c == '"')
                         break;
 
-                if (c == '\\') {
+                if (c == '&') {
                         c = fgetc(f);
                         if (c == EOF) {
@@ -735 +763 @@
                                 goto error;
                         }
+
+                        if (c == 'q') {
+                                rc = sif_get_verify_chars(f, "uot;");
+                                if (rc != EOK)
+                                        goto error;
+                        } else if (c == 'l') {
+                                rc = sif_get_verify_chars(f, "t;");
+                                if (rc != EOK)
+                                        goto error;
+                        } else {
+                                rc = EIO;
+                                goto error;
+                        }
                 }
 
                 if (sidx >= str_size) {
                         str_size *= 2;
-                        str = realloc(str, str_size + 1);
-                        if (str == NULL) {
+                        nstr = realloc(str, str_size + 1);
+                        if (nstr == NULL) {
                                 rc = ENOMEM;
                                 goto error;
                         }
+
+                        str = nstr;
                 }
 
@@ -772 +815 @@
         int c;
 
-        rc = sif_import_string(f, &aname);
+        rc = sif_import_name(f, &aname);
         if (rc != EOK)
                 goto error;
@@ -815 +858 @@
         errno_t rc;
 
-        rc = sif_export_string(attr->aname, f);
+        rc = sif_export_name(attr->aname, f);
         if (rc != EOK)
                 return rc;
@@ -841 +884 @@
         sif_node_t *child;
 
-        rc = sif_export_string(node->ntype, f);
+        if (fputc('<', f) == EOF)
+                return EIO;
+
+        rc = sif_export_name(node->ntype, f);
         if (rc != EOK)
                 return rc;
@@ -847 +893 @@
         /* Attributes */
 
-        if (fputc('(', f) == EOF)
-                return EIO;
-
         attr = sif_node_first_attr(node);
         while (attr != NULL) {
+                if (fputc(' ', f) == EOF)
+                        return EIO;
+
                 rc = sif_export_attr(attr, f);
                 if (rc != EOK)
@@ -859 +905 @@
         }
 
-        if (fputc(')', f) == EOF)
+        if (fputs(">\n", f) == EOF)
                 return EIO;
 
         /* Child nodes */
-
-        if (fputc('{', f) == EOF)
-                return EIO;
 
         child = sif_node_first_child(node);
@@ -876 +919 @@
         }
 
-        if (fputc('}', f) == EOF)
+        if (fputs("</", f) == EOF)
+                return EIO;
+
+        rc = sif_export_name(node->ntype, f);
+        if (rc != EOK)
+                return rc;
+
+        if (fputs(">\n", f) == EOF)
                 return EIO;
 
@@ -887 +937 @@
  * @param f File
  * @param rnode Place to store pointer to imported node
+ * @param rendtag Place to store @c true iff end tag is encountered
  * @return EOK on success, EIO on I/O error
  */
-static errno_t sif_import_node(sif_node_t *parent, FILE *f, sif_node_t **rnode)
+static errno_t sif_import_node(sif_node_t *parent, FILE *f, sif_node_t **rnode,
+    bool *rendtag)
 {
         errno_t rc;
@@ -895 +947 @@
         sif_node_t *child;
         sif_attr_t *attr = NULL;
+        bool endtag;
+        bool cendtag;
         char *ntype;
         int c;
@@ -902 +956 @@
                 return ENOMEM;
 
-        rc = sif_import_string(f, &ntype);
+        c = fgetc(f);
+        while (isspace(c))
+                c = fgetc(f);
+
+        if (c != '<') {
+                rc = EIO;
+                goto error;
+        }
+
+        c = fgetc(f);
+        if (c == '/') {
+                endtag = true;
+        } else {
+                endtag = false;
+                ungetc(c, f);
+        }
+
+        rc = sif_import_name(f, &ntype);
         if (rc != EOK)
                 goto error;
@@ -909 +980 @@
 
         /* Attributes */
-
-        c = fgetc(f);
-        if (c != '(') {
-                rc = EIO;
-                goto error;
-        }
 
         c = fgetc(f);
@@ -922 +987 @@
         }
 
-        while (c != ')') {
+        while (c != '>') {
+                /* End tags cannot have attributes */
+                if (endtag) {
+                        rc = EIO;
+                        goto error;
+                }
+
+                while (isspace(c))
+                        c = fgetc(f);
                 ungetc(c, f);
 
@@ -940 +1013 @@
         /* Child nodes */
 
-        c = fgetc(f);
-        if (c != '{') {
-                rc = EIO;
-                goto error;
-        }
-
-        c = fgetc(f);
-        if (c == EOF) {
-                rc = EIO;
-                goto error;
-        }
-
-        while (c != '}') {
-                ungetc(c, f);
-
-                rc = sif_import_node(node, f, &child);
-                if (rc != EOK)
-                        goto error;
-
-                list_append(&child->lparent, &node->children);
-
-                c = fgetc(f);
-                if (c == EOF) {
-                        rc = EIO;
-                        goto error;
+        if (!endtag) {
+                while (true) {
+                        rc = sif_import_node(node, f, &child, &cendtag);
+                        if (rc != EOK)
+                                goto error;
+
+                        if (cendtag) {
+                                sif_node_delete(child);
+                                break;
+                        }
+
+                        list_append(&child->lparent, &node->children);
                 }
         }
 
         *rnode = node;
+        *rendtag = endtag;
         return EOK;
 error: