From b08db17162fddda97e6ee1ac625eae1430d22b64 Mon Sep 17 00:00:00 2001
From: Aleksey Veresov <aleksey@veresov.pro>
Date: Fri, 29 Nov 2019 18:50:04 +0300
Subject: [magi] Almost done!

---
 src/multipart.c | 478 ++++++++++++++++++++++++++------------------------------
 1 file changed, 220 insertions(+), 258 deletions(-)

(limited to 'src/multipart.c')

diff --git a/src/multipart.c b/src/multipart.c
index c3e4a2d..5d7f261 100644
--- a/src/multipart.c
+++ b/src/multipart.c
@@ -3,6 +3,7 @@
 
 #include "error.h"
 #include "param.h"
+#include "utils.h"
 #include <ctype.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -43,114 +44,128 @@ enum st {
 };
 
 struct automata {
-    struct magi_field_list ** list;
-    struct magi_field         field;
-    struct magi_param         param;
-    char *                    buf;
-    int                       buf_size;
-    int                       size;
-    int                       len;
-    char *                    boundary;
-    int                       boundary_pos;
-    int                       boundary_len;
-    int                       is_end_suspected;
-    int                       is_CR_readed;
-    int                       is_quoted;
-    void (*callback)(struct magi_field * field, char * buffer, int size);
+    struct magi_request * request;
+    struct magi_file      file;
+    struct magi_param     param;
+    struct magi_param     subparam;
+    char *                buf;
+    int                   buf_size;
+    int                   size;
+    int                   len;
+    char *                boundary;
+    int                   boundary_pos;
+    int                   boundary_len;
+    int                   is_end_suspected;
+    int                   is_CR_readed;
+    int                   is_quoted;
 };
 
 
 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
  * Automata Shortcuts
  */
-static int content_disposition(struct automata * a)
+static char * extract_filename(char * n)
 {
-    int    ok   = 1;
-    char * name = strchr(a->param.data, '=');
-    if (name) {
-        name += strspn(name, " \t") + 1;
-        if (*name == '"') {
-            ++name;
-            a->field.name = create_str(name, strchr(name, '"'));
-            if (a->field.name == 0) {
-                ok = 0;
-            } else if (a->field.name[0] == 0) {
-                ok = 0;
-                magi_error_set(
-                    "[multipart] Wrong content-disposition quotation.");
-            }
-        } else {
-            a->field.name = create_str(name, name + strcspn(name, " \t"));
-            if (a->field.name == 0) {
-                ok = 0;
-            } else if (!is_str_token(a->field.name)) {
-                ok = 0;
-                magi_error_set(
-                    "[multipart] Content-disposition value is not valid, "
-                    "readed: %s.",
-                    a->field.name);
-            }
-        }
-        if (ok) {
-            free(a->param.name);
-            free(a->param.data);
-            a->param.name = 0;
-            a->param.data = 0;
-        }
+    n = strchr(n, '=');
+    if (!n) {
+        return 0;
+    }
+    n += strspn(n, " \t") + 1;
+    if (*n == '"') {
+        ++n;
+        return magi_str_create_copy(n, strchr(n, '"'));
     } else {
-        ok = 0;
-        magi_error_set("[multipart] Content-disposition has no '=' symbol.");
+        return magi_str_create_copy(n, n + strcspn(n, " \t"));
     }
-    return ok;
 }
 
-static int param_end(struct automata * a)
+static int content_disposition(struct automata * a)
 {
-    int ok = 1;
-    lowercase(a->param.name);
-    if (!strcmp(a->param.name, "content-disposition")) {
-        ok = content_disposition(a);
+    char * n = strchr(a->subparam.data, '=');
+    if (!n) {
+        return 0;
+    }
+    n += strspn(n, " \t") + 1;
+    if (*n == '"') {
+        ++n;
+        a->param.name = magi_str_create_copy(n, strchr(n, '"'));
+        if (!a->param.name || !*a->param.name) {
+            return 0;
+        }
     } else {
-        ok            = magi_param_list_add(&a->field.params, &a->param);
-        a->param.name = 0;
-        a->param.data = 0;
+        a->param.name = magi_str_create_copy(n, n + strcspn(n, " \t"));
+        if (!a->param.name || !is_str_token(a->param.name)) {
+            return 0;
+        }
+    }
+    a->file.file_name = extract_filename(n);
+    if (a->file.file_name) {
+        a->file.param_name = a->param.name;
+        a->file.params     = 0;
     }
+    free(a->subparam.name);
+    free(a->subparam.data);
+    a->subparam.name = 0;
+    a->subparam.data = 0;
+    return 1;
+}
+
+static int subparam_end(struct automata * a)
+{
     a->size = 1;
     a->len  = 0;
-    return ok;
+    magi_str_lowercase(a->subparam.name);
+    if (!strcmp(a->subparam.name, "content-disposition")) {
+        return content_disposition(a);
+    } else if (magi_param_list_add(&a->file.params, &a->subparam)) {
+        a->subparam.name = 0;
+        a->subparam.data = 0;
+        return 1;
+    }
+    return 0;
 }
 
-static int field_end(struct automata * a)
+static int param_end(struct automata * a)
 {
-    int ok;
-    if (a->field.name == 0) {
-        ok = 0;
-        magi_error_set("[multipart] Field name is empty or not specified.");
-    } else {
-        if (a->callback) {
-            a->callback(&a->field, a->buf, a->buf_size);
-            a->buf_size = 0;
-        }
-        a->field.len = a->len;
-        ok           = magi_field_list_add(a->list, &a->field);
-        if (!ok) {
-            free(a->field.name);
-            free(a->field.data);
-            magi_param_list_destroy(a->field.params);
-            free(a->field.params);
+    if (!a->param.name) {
+        a->request->error = magi_error_multipart;
+        return 0;
+    }
+    if (a->file.file_name) {
+        a->request->file_callback(&a->file, a->buf, a->buf_size, 1,
+                                  a->request->file_callback_userdata);
+        a->buf_size = 0;
+        if (!magi_file_list_add(&a->request->files, &a->file)) {
+            free(a->file.file_name);
+            free(a->file.param_name);
+            magi_param_list_destroy(a->file.params);
+            free(a->file.params);
+            a->request->error = magi_error_multipart;
+            return 0;
         }
-        a->field.name   = 0;
-        a->field.data   = 0;
-        a->field.params = 0;
-        a->size         = 1;
-        a->len          = 0;
+        a->file.file_name  = 0;
+        a->file.param_name = 0;
+        a->file.params     = 0;
+        a->size            = 1;
+        a->len             = 0;
+        return 1;
+    }
+    if (!magi_param_list_add(&a->request->params, &a->param)) {
+        free(a->param.name);
+        free(a->param.data);
+        a->request->error = magi_error_multipart;
+        return 0;
     }
-    return ok;
+    a->param.name = 0;
+    a->param.data = 0;
+    a->size       = 1;
+    a->len        = 0;
+    return 1;
 }
 
 
 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
- * Boundary interfaces
+ * Boundary Interfaces
  */
 static char sepget(const struct automata * a)
 {
@@ -214,214 +229,169 @@ static int is_semiend(const struct automata * a)
 }
 
 
+/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
+ * Data Reading
+ */
+static void apply_callback(struct automata * a)
+{
+    int full = a->buf_size == a->request->file_callback_addon_max;
+    if (a->file.file_name && full) {
+        a->request->file_callback(&a->file, a->buf, a->buf_size, 0,
+                                  a->request->file_callback_userdata);
+        a->buf_size = 0;
+    }
+}
+
+static enum st data_add_act(
+    struct automata * a, char c, char ** dest, int * len, int * size)
+{
+    int pos = a->boundary_pos;
+    for (a->boundary_pos = 0; a->boundary_pos < pos; ++a->boundary_pos) {
+        if (a->is_end_suspected) {
+            magi_str_add(dest, len, size, endget(a));
+        } else {
+            magi_str_add(dest, len, size, sepget(a));
+        }
+        apply_callback(a);
+    }
+    a->boundary_pos     = 0;
+    a->is_end_suspected = 0;
+
+    if (sepget(a) != c) {
+        magi_str_add(dest, len, size, c);
+        apply_callback(a);
+        return st_data;
+    } else {
+        a->boundary_pos++;
+        if (a->boundary_pos == seplen(a)) {
+            param_end(a);
+            return st_pname_pre;
+        }
+        return st_data;
+    }
+}
+
+static enum st data_add(struct automata * a, char c)
+{
+    if (a->file.file_name) {
+        int max = a->request->file_callback_addon_max + 1;
+        return data_add_act(a, c, &a->buf, &a->buf_size, &max);
+    } else {
+        return data_add_act(a, c, &a->param.data, &a->len, &a->size);
+    }
+}
+
+
 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
  * State analysers
  */
 static enum st parse_begin(struct automata * a, char c)
 {
-    enum st state;
-    if (sepget(a) == c) {
+    if (sepget(a) != c) {    /* 'c' is not wanted character from separator; */
+        a->boundary_pos = 0; /* so nullify progress in reading separator. */
+    } else {
         a->boundary_pos++;
         if (a->boundary_pos == seplen(a)) {
-            state = st_pname_pre;
-        } else {
-            state = st_begin;
+            return st_pname_pre; /* Separator is completed, so move on. */
         }
-    } else {
-        state           = st_begin;
-        a->boundary_pos = 0;
     }
-    return state;
+    return st_begin;
 }
 
 static enum st parse_pname_pre(struct automata * a, char c)
 {
-    enum st state;
     if (a->is_CR_readed) {
-        a->is_CR_readed = 0;
-        if (c == '\n') {
-            state           = st_data;
-            a->boundary_pos = 0;
-        } else {
-            state = st_error;
-            magi_error_set(
-                "[multipart] Waiting for name, CR is readed alone.");
+        if (c != '\n') {
+            return st_error;
         }
+        a->is_CR_readed = 0;
+        a->boundary_pos = 0;
+        return st_data;
     } else if (c == '\r') {
-        state           = st_pname_pre;
         a->is_CR_readed = 1;
-    } else if (is_token(c)) {
-        if (add(&a->param.name, &a->len, &a->size, c)) {
-            state = st_pname;
-        } else {
-            state = st_error;
-        }
-    } else {
-        state = st_error;
-        magi_error_set(
-            "[multipart] Waiting for name, readed: \\%o (render: %c).", c, c);
+        return st_pname_pre;
+    } else if (is_token(c) &&
+               magi_str_add(&a->subparam.name, &a->len, &a->size, c)) {
+        return st_pname;
     }
-    return state;
+    return st_error;
 }
 
 static enum st parse_pname(struct automata * a, char c)
 {
-    enum st state;
     if (c == ':') {
-        state   = st_pdata;
         a->len  = 0;
         a->size = 1;
+        return st_pdata;
     } else if (c == ' ' || c == '\t') {
-        state = st_pname_end;
-    } else if (is_token(c)) {
-        if (add(&a->param.name, &a->len, &a->size, c)) {
-            state = st_pname;
-        } else {
-            state = st_error;
-        }
-    } else {
-        state = st_error;
-        magi_error_set("[multipart] Reading name, readed: \\%o (render: %c).",
-                       c, c);
+        return st_pname_end;
+    } else if (is_token(c) &&
+               magi_str_add(&a->subparam.name, &a->len, &a->size, c)) {
+        return st_pname;
     }
-    return state;
+    return st_error;
 }
 
 static enum st parse_pname_end(struct automata * a, char c)
 {
-    enum st state;
     if (c == ':') {
-        state   = st_pdata;
         a->len  = 0;
         a->size = 1;
+        return st_pdata;
     } else if (c == ' ' || c == '\t') {
-        state = st_pname_end;
-    } else {
-        state = st_error;
-        magi_error_set("[multipart] Waiting for name-value separator, "
-                       "readed: \\%o (render: %c).",
-                       c, c);
+        return st_pname_end;
     }
-    return state;
+    return st_error;
 }
 
 static enum st parse_pdata(struct automata * a, char c)
 {
-    enum st state;
     if (a->is_CR_readed) {
         a->is_CR_readed = 0;
         if (c == '\n') {
-            if (param_end(a)) {
-                state = st_pname_pre;
-            } else {
-                state = st_error;
+            if (subparam_end(a)) {
+                return st_pname_pre;
             }
-        } else if (add(&a->param.data, &a->len, &a->size, '\r')) {
-            if (add(&a->param.data, &a->len, &a->size, c)) {
-                state = st_pdata;
-            } else {
-                state = st_error;
-            }
-        } else {
-            state = st_error;
+            return st_error;
+        } else if (magi_str_add(&a->subparam.data, &a->len, &a->size, '\r') &&
+                   magi_str_add(&a->subparam.data, &a->len, &a->size, c)) {
+            return st_pdata;
         }
+        return st_error;
     } else if (c == '\r') {
-        state           = st_pdata;
         a->is_CR_readed = 1;
-    } else {
-        if (add(&a->param.data, &a->len, &a->size, c)) {
-            state = st_pdata;
-        } else {
-            state = st_error;
-        }
+        return st_pdata;
+    } else if (magi_str_add(&a->subparam.data, &a->len, &a->size, c)) {
+        return st_pdata;
     }
-    return state;
-}
-
-static void apply_callback(struct automata * a)
-{
-    if (a->callback && a->buf_size == magi_parse_multipart_callback_size) {
-        a->callback(&a->field, a->buf, a->buf_size);
-        a->buf_size = 0;
-    }
-}
-
-static enum st data_add(struct automata * a, char c)
-{
-    static int max_buf_size = magi_parse_multipart_callback_size + 1;
-    enum st    state;
-    char **    dest;
-    int *      len;
-    int *      size;
-    int        pos  = a->boundary_pos;
-    state           = st_data;
-    a->boundary_pos = 0;
-    if (a->callback) {
-        dest = &a->buf;
-        len  = &a->buf_size;
-        size = &max_buf_size;
-    } else {
-        dest = &a->field.data;
-        len  = &a->len;
-        size = &a->size;
-    }
-    while (a->boundary_pos < pos) {
-        if (a->is_end_suspected) {
-            add(dest, len, size, endget(a));
-        } else {
-            add(dest, len, size, sepget(a));
-        }
-        apply_callback(a);
-        a->boundary_pos++;
-    }
-    a->boundary_pos     = 0;
-    a->is_end_suspected = 0;
-    if (sepget(a) == c) {
-        a->boundary_pos++;
-        if (a->boundary_pos == seplen(a)) {
-            state = st_pname_pre;
-            field_end(a);
-        } else {
-            state = st_data;
-        }
-    } else {
-        add(dest, len, size, c);
-        apply_callback(a);
-    }
-    return state;
+    return st_error;
 }
 
 static enum st parse_data(struct automata * a, char c)
 {
-    enum st state;
     if (a->is_end_suspected) {
-        if (endget(a) == c) {
-            a->boundary_pos++;
-            if (a->boundary_pos == endlen(a)) {
-                state = st_end;
-                field_end(a);
-            } else {
-                state = st_data;
-            }
-        } else {
-            state = data_add(a, c);
+        if (endget(a) != c) {
+            return data_add(a, c);
+        }
+        a->boundary_pos++;
+        if (a->boundary_pos == endlen(a)) {
+            param_end(a);
+            return st_end;
         }
+        return st_data;
     } else if (sepget(a) == c) {
         a->boundary_pos++;
         if (a->boundary_pos == seplen(a)) {
-            state = st_pname_pre;
-            field_end(a);
-        } else {
-            state = st_data;
+            param_end(a);
+            return st_pname_pre;
         }
+        return st_data;
     } else if ((a->boundary_pos == seplen(a) - 2) && endget(a) == c) {
-        state               = st_data;
         a->is_end_suspected = 1;
         a->boundary_pos++;
-    } else {
-        state = data_add(a, c);
+        return st_data;
     }
-    return state;
+    return data_add(a, c);
 }
 
 static enum st parse_end(struct automata * a, char c)
@@ -433,14 +403,13 @@ static enum st parse_end(struct automata * a, char c)
 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
  * Automata Runner
  */
-static int run_automata(struct automata * a,
-                        int (*next)(void * thing),
-                        void * thing)
+static void run_automata(struct automata * a,
+                         int (*next)(void * next_userdata),
+                         void * next_userdata)
 {
-    int     ok    = 1;
     enum st state = st_begin;
-    int     c;
-    for (c = next(thing); state && c != EOF; c = next(thing)) {
+    int     c     = next(next_userdata);
+    while (state && state != st_end && c != EOF) {
         switch (state) {
         case st_begin:
             state = parse_begin(a, c);
@@ -465,43 +434,36 @@ static int run_automata(struct automata * a,
         default:
             break;
         }
+        c = next(next_userdata);
     }
     if (state == st_data && is_semiend(a)) {
         state = st_end;
-        field_end(a);
+        param_end(a);
     }
     if (state != st_end) {
-        ok = 0;
-        if (state != st_error) {
-            magi_error_set("[multipart] Input ended unexpectedly.");
-        }
-        free(a->field.name);
-        free(a->field.data);
+        a->request->error = magi_error_multipart;
+        free(a->subparam.name);
+        free(a->subparam.data);
     }
-    return ok;
 }
 
 
 /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
  * Automata Interfaces
  */
-int magi_parse_multipart(struct magi_request * request,
-                         char *                boundary,
-                         int (*get_next)(void * userdata),
-                         void * get_next_userdata)
+void magi_multipart(struct magi_request * request,
+                    char *                boundary,
+                    int (*next)(void * userdata),
+                    void * next_userdata)
 {
-    struct automata a = {
-        0, { 0, 0, 0 }, { 0, 0 }, 0, 0, 1, 0, 0, 2, 0, 0, 0
-    };
-    int ok         = 0;
-    a.list         = list;
-    a.boundary     = boundary;
-    a.boundary_len = strlen(boundary);
-    a.callback     = callback;
-    a.buf          = malloc(magi_parse_multipart_callback_size + 1);
+    struct automata a = { 0, { 0, 0, 0 }, { 0, 0 }, { 0, 0 }, 0, 0, 1,
+                          0, 0,           2,        0,        0, 0 };
+    a.request         = request;
+    a.boundary        = boundary;
+    a.boundary_len    = strlen(boundary);
+    a.buf             = malloc(request->file_callback_addon_max + 1);
     if (a.buf) {
-        ok = run_automata(&a, get_next, get_next_arg);
+        run_automata(&a, next, next_userdata);
         free(a.buf);
     }
-    return ok;
 }
-- 
cgit v1.2.3