Initial support for blobs and upload

2 files changed, 307 insertions, 42 deletions

diff --git a/src/c/driver.c b/src/c/driver.c
index f7456ed9..fa4f474b 100644
--- a/src/c/driver.c
+++ b/src/c/driver.c
@@ -1,5 +1,6 @@
-#include <stdio.h>
+#define _GNU_SOURCE
 
+#include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
 #include <sys/types.h>
@@ -147,9 +148,11 @@ void uw_sign(const char *in, char *out) {
 static void *worker(void *data) {
   int me = *(int *)data, retries_left = MAX_RETRIES;
   uw_context ctx = new_context();
+  size_t buf_size = 1;
+  char *buf = malloc(buf_size);
 
   while (1) {
-    char buf[uw_bufsize+1], *back = buf, *s, *post;
+    char *back = buf, *s, *post;
     int sock, dont_close = 0;
 
     pthread_mutex_lock(&queue_mutex);
@@ -162,7 +165,17 @@ static void *worker(void *data) {
 
     while (1) {
       unsigned retries_left = MAX_RETRIES;
-      int r = recv(sock, back, uw_bufsize - (back - buf), 0);
+      int r;
+
+      if (back - buf == buf_size) {
+        char *new_buf;
+        buf_size *= 2;
+        new_buf = realloc(buf, buf_size);
+        back = new_buf + (back - buf);
+        buf = new_buf;
+      }
+
+      r = recv(sock, back, buf_size - (back - buf), 0);
 
       if (r < 0) {
         fprintf(stderr, "Recv failed\n");
@@ -182,8 +195,12 @@ static void *worker(void *data) {
       if (s = strstr(buf, "\r\n\r\n")) {
         failure_kind fk;
         int is_post = 0;
+        char *boundary = NULL;
+        size_t boundary_len;
         char *cmd, *path, *headers, path_copy[uw_bufsize+1], *inputs, *after_headers;
 
+        //printf("All: %s\n", buf);
+
         s[2] = 0;
         after_headers = s + 4;
 
@@ -196,7 +213,7 @@ static void *worker(void *data) {
         headers = s + 2;
         cmd = s = buf;
 
-        printf("Read: %s\n", buf);
+        //printf("Read: %s\n", buf);
       
         if (!strsep(&s, " ")) {
           fprintf(stderr, "No first space in HTTP command\n");
@@ -208,17 +225,25 @@ static void *worker(void *data) {
         if (!strcmp(cmd, "POST")) {
           char *clen_s = uw_Basis_requestHeader(ctx, "Content-length");
           if (!clen_s) {
-            printf("No Content-length with POST\n");
+            fprintf(stderr, "No Content-length with POST\n");
             goto done;
           }
           int clen = atoi(clen_s);
           if (clen < 0) {
-            printf("Negative Content-length with POST\n");
+            fprintf(stderr, "Negative Content-length with POST\n");
             goto done;
           }
 
           while (back - after_headers < clen) {
-            r = recv(sock, back, uw_bufsize - (back - buf), 0);
+            if (back - buf == buf_size) {
+              char *new_buf;
+              buf_size *= 2;
+              new_buf = realloc(buf, buf_size);
+              back = new_buf + (back - buf);
+              buf = new_buf;
+            }
+
+            r = recv(sock, back, buf_size - (back - buf), 0);
 
             if (r < 0) {
               fprintf(stderr, "Recv failed\n");
@@ -235,6 +260,19 @@ static void *worker(void *data) {
           }
 
           is_post = 1;
+
+          clen_s = uw_Basis_requestHeader(ctx, "Content-type");
+          if (clen_s && !strncasecmp(clen_s, "multipart/form-data", 19)) {
+            if (strncasecmp(clen_s + 19, "; boundary=", 11)) {
+              fprintf(stderr, "Bad multipart boundary spec");
+              break;
+            }
+
+            boundary = clen_s + 28;
+            boundary[0] = '-';
+            boundary[1] = '-';
+            boundary_len = strlen(boundary);
+          }
         } else if (strcmp(cmd, "GET")) {
           fprintf(stderr, "Not ready for non-GET/POST command: %s\n", cmd);
           break;
@@ -262,26 +300,134 @@ static void *worker(void *data) {
           break;
         }
 
-        if (is_post)
-          inputs = after_headers;
-        else if (inputs = strchr(path, '?'))
-          *inputs++ = 0;
-        if (inputs) {
-          char *name, *value;
-
-          while (*inputs) {
-            name = inputs;
-            if (inputs = strchr(inputs, '&'))
-              *inputs++ = 0;
-            else
-              inputs = strchr(name, 0);
-
-            if (value = strchr(name, '=')) {
-              *value++ = 0;
-              uw_set_input(ctx, name, value);
+        if (boundary) {
+          char *part = after_headers, *after_sub_headers, *header, *after_header;
+          size_t part_len;
+
+          part = strstr(part, boundary);
+          if (!part) {
+            fprintf(stderr, "Missing first multipart boundary\n");
+            break;
+          }
+          part += boundary_len;
+
+          while (1) {
+            char *name = NULL, *filename = NULL, *type = NULL;
+
+            if (part[0] == '-' && part[1] == '-')
+              break;
+
+            if (*part != '\r') {
+              fprintf(stderr, "No \\r after multipart boundary\n");
+              goto done;
+            }
+            ++part;
+            if (*part != '\n') {
+              fprintf(stderr, "No \\n after multipart boundary\n");
+              goto done;
+            }
+            ++part;
+            
+            if (!(after_sub_headers = strstr(part, "\r\n\r\n"))) {
+              fprintf(stderr, "Missing end of headers after multipart boundary\n");
+              goto done;
+            }
+            after_sub_headers[2] = 0;
+            after_sub_headers += 4;
+
+            for (header = part; after_header = strstr(header, "\r\n"); header = after_header + 2) {
+              char *colon, *after_colon;
+
+              *after_header = 0;
+              if (!(colon = strchr(header, ':'))) {
+                fprintf(stderr, "Missing colon in multipart sub-header\n");
+                goto done;
+              }
+              *colon++ = 0;
+              if (*colon++ != ' ') {
+                fprintf(stderr, "No space after colon in multipart sub-header\n");
+                goto done;
+              }
+
+              if (!strcasecmp(header, "Content-Disposition")) {
+                if (strncmp(colon, "form-data; ", 11)) {
+                  fprintf(stderr, "Multipart data is not \"form-data\"\n");
+                  goto done;
+                }
+
+                for (colon += 11; after_colon = strchr(colon, '='); colon = after_colon) {
+                  char *data;
+                  after_colon[0] = 0;
+                  if (after_colon[1] != '"') {
+                    fprintf(stderr, "Disposition setting is missing initial quote\n");
+                    goto done;
+                  }
+                  data = after_colon+2;
+                  if (!(after_colon = strchr(data, '"'))) {
+                    fprintf(stderr, "Disposition setting is missing final quote\n");
+                    goto done;
+                  }
+                  after_colon[0] = 0;
+                  ++after_colon;
+                  if (after_colon[0] == ';' && after_colon[1] == ' ')
+                    after_colon += 2;
+
+                  if (!strcasecmp(colon, "name"))
+                    name = data;
+                  else if (!strcasecmp(colon, "filename"))
+                    filename = data;
+                }
+              } else if (!strcasecmp(header, "Content-Type")) {
+                type = colon;
+              }
+            }
+
+            part = memmem(after_sub_headers, back - after_sub_headers, boundary, boundary_len);
+            if (!part) {
+              fprintf(stderr, "Missing boundary after multipart payload\n");
+              goto done;
+            }
+            part[-2] = 0;
+            part_len = part - after_sub_headers - 2;
+            part[0] = 0;
+            part += boundary_len;
+
+            if (filename) {
+              uw_Basis_file *f = malloc(sizeof(uw_Basis_file));
+              uw_Basis_files fs = { 1, f };
+
+              f->name = filename;
+              f->data.size = part_len;
+              f->data.data = after_sub_headers;
+
+              uw_set_file_input(ctx, name, fs);
+            } else
+              uw_set_input(ctx, name, after_sub_headers);
+          }
+        }
+        else {
+          if (is_post)
+            inputs = after_headers;
+          else if (inputs = strchr(path, '?'))
+            *inputs++ = 0;
+
+          if (inputs) {
+            char *name, *value;
+
+            while (*inputs) {
+              name = inputs;
+              if (inputs = strchr(inputs, '&'))
+                *inputs++ = 0;
+              else
+                inputs = strchr(name, 0);
+
+              if (value = strchr(name, '=')) {
+                *value++ = 0;
+                uw_set_input(ctx, name, value);
+              }
+              else
+                uw_set_input(ctx, name, "");
             }
-            else
-              uw_set_input(ctx, name, "");
           }
         }
 
diff --git a/src/c/urweb.c b/src/c/urweb.c
index 6266e12d..99564605 100644
--- a/src/c/urweb.c
+++ b/src/c/urweb.c
@@ -282,11 +282,23 @@ typedef struct {
   buf msgs;
 } delta;
 
+typedef enum {
+  UNSET, NORMAL, FILES
+} input_kind;
+
+typedef struct {
+  input_kind kind;
+  union {
+    char *normal;
+    uw_Basis_files files;
+  } data;
+} input;
+
 struct uw_context {
   char *headers, *headers_end;
 
   buf outHeaders, page, heap, script;
-  char **inputs;
+  input *inputs;
 
   int source_count;
 
@@ -325,7 +337,7 @@ uw_context uw_init() {
   buf_init(&ctx->script, 1);
   ctx->script.start[0] = 0;
 
-  ctx->inputs = calloc(uw_inputs_len, sizeof(char *));
+  ctx->inputs = calloc(uw_inputs_len, sizeof(input));
 
   ctx->db = NULL;
 
@@ -398,7 +410,7 @@ void uw_reset_keep_request(uw_context ctx) {
 
 void uw_reset(uw_context ctx) {
   uw_reset_keep_request(ctx);
-  memset(ctx->inputs, 0, uw_inputs_len * sizeof(char *));
+  memset(ctx->inputs, 0, uw_inputs_len * sizeof(input));
 }
 
 void uw_db_init(uw_context);
@@ -544,9 +556,9 @@ char *uw_error_message(uw_context ctx) {
   return ctx->error_message;
 }
 
-int uw_input_num(char*);
+extern int uw_input_num(const char*);
 
-void uw_set_input(uw_context ctx, char *name, char *value) {
+void uw_set_input(uw_context ctx, const char *name, char *value) {
   int n = uw_input_num(name);
 
   if (n < 0)
@@ -555,9 +567,8 @@ void uw_set_input(uw_context ctx, char *name, char *value) {
   if (n >= uw_inputs_len)
     uw_error(ctx, FATAL, "For input name %s, index %d is out of range", name, n);
 
-  ctx->inputs[n] = value;
-
-  //printf("[%d] %s = %s\n", n, name, value);
+  ctx->inputs[n].kind = NORMAL;
+  ctx->inputs[n].data.normal = value;
 }
 
 char *uw_get_input(uw_context ctx, int n) {
@@ -565,8 +576,17 @@ char *uw_get_input(uw_context ctx, int n) {
     uw_error(ctx, FATAL, "Negative input index %d", n);
   if (n >= uw_inputs_len)
     uw_error(ctx, FATAL, "Out-of-bounds input index %d", n);
-  //printf("[%d] = %s\n", n, ctx->inputs[n]);
-  return ctx->inputs[n];
+
+  switch (ctx->inputs[n].kind) {
+  case UNSET:
+    return NULL;
+  case FILES:
+    uw_error(ctx, FATAL, "Tried to read a files form input as normal");
+  case NORMAL:
+    return ctx->inputs[n].data.normal;
+  default:
+    uw_error(ctx, FATAL, "Impossible input kind");
+  }
 }
 
 char *uw_get_optional_input(uw_context ctx, int n) {
@@ -574,8 +594,51 @@ char *uw_get_optional_input(uw_context ctx, int n) {
     uw_error(ctx, FATAL, "Negative input index %d", n);
   if (n >= uw_inputs_len)
     uw_error(ctx, FATAL, "Out-of-bounds input index %d", n);
-  //printf("[%d] = %s\n", n, ctx->inputs[n]);
-  return (ctx->inputs[n] == NULL ? "" : ctx->inputs[n]);
+
+  switch (ctx->inputs[n].kind) {
+  case UNSET:
+    return "";
+  case FILES:
+    uw_error(ctx, FATAL, "Tried to read a files form input as normal");
+  case NORMAL:
+    return ctx->inputs[n].data.normal;
+  default:
+    uw_error(ctx, FATAL, "Impossible input kind");
+  }
+}
+
+void uw_set_file_input(uw_context ctx, const char *name, uw_Basis_files fs) {
+  int n = uw_input_num(name);
+
+  if (n < 0)
+    uw_error(ctx, FATAL, "Bad file input name %s", name);
+
+  if (n >= uw_inputs_len)
+    uw_error(ctx, FATAL, "For file input name %s, index %d is out of range", name, n);
+
+  ctx->inputs[n].kind = FILES;
+  ctx->inputs[n].data.files = fs;
+}
+
+uw_Basis_files uw_get_file_input(uw_context ctx, int n) {
+  if (n < 0)
+    uw_error(ctx, FATAL, "Negative file input index %d", n);
+  if (n >= uw_inputs_len)
+    uw_error(ctx, FATAL, "Out-of-bounds file input index %d", n);
+
+  switch (ctx->inputs[n].kind) {
+  case UNSET:
+    {
+      uw_Basis_files fs = {};
+      return fs;
+    }
+  case FILES:
+    return ctx->inputs[n].data.files;
+  case NORMAL:
+    uw_error(ctx, FATAL, "Tried to read a normal form input as files");
+  default:
+    uw_error(ctx, FATAL, "Impossible input kind");
+  }
 }
 
 void uw_set_script_header(uw_context ctx, const char *s) {
@@ -1393,7 +1456,7 @@ uw_Basis_string uw_Basis_strcat(uw_context ctx, uw_Basis_string s1, uw_Basis_str
   return s;
 }
 
-uw_Basis_string uw_Basis_strdup(uw_context ctx, uw_Basis_string s1) {
+uw_Basis_string uw_strdup(uw_context ctx, uw_Basis_string s1) {
   int len = strlen(s1) + 1;
   char *s;
 
@@ -1407,9 +1470,9 @@ uw_Basis_string uw_Basis_strdup(uw_context ctx, uw_Basis_string s1) {
   return s;
 }
 
-uw_Basis_string uw_Basis_maybe_strdup(uw_context ctx, uw_Basis_string s1) {
+uw_Basis_string uw_maybe_strdup(uw_context ctx, uw_Basis_string s1) {
   if (s1)
-    return uw_Basis_strdup(ctx, s1);
+    return uw_strdup(ctx, s1);
   else
     return NULL;
 }
@@ -1477,7 +1540,7 @@ uw_Basis_string uw_Basis_sqlifyString(uw_context ctx, uw_Basis_string s) {
       if (isprint(c))
         *s2++ = c;
       else {
-        sprintf(s2, "\\%3o", c);
+        sprintf(s2, "\\%03o", c);
         s2 += 4;
       }
     }
@@ -1488,6 +1551,43 @@ uw_Basis_string uw_Basis_sqlifyString(uw_context ctx, uw_Basis_string s) {
   return r;
 }
 
+uw_Basis_string uw_Basis_sqlifyBlob(uw_context ctx, uw_Basis_blob b) {
+  char *r, *s2;
+  size_t i;
+
+  uw_check_heap(ctx, b.size * 5 + 11);
+
+  r = s2 = ctx->heap.front;
+  *s2++ = 'E';
+  *s2++ = '\'';
+
+  for (i = 0; i < b.size; ++i) {
+    char c = b.data[i];
+
+    switch (c) {
+    case '\'':
+      strcpy(s2, "\\'");
+      s2 += 2;
+      break;
+    case '\\':
+      strcpy(s2, "\\\\\\\\");
+      s2 += 4;
+      break;
+    default:
+      if (isprint(c))
+        *s2++ = c;
+      else {
+        sprintf(s2, "\\\\%03o", c);
+        s2 += 5;
+      }
+    }
+  }
+
+  strcpy(s2, "'::bytea");
+  ctx->heap.front = s2 + 9;
+  return r;
+}
+
 char *uw_Basis_sqlifyChannel(uw_context ctx, uw_Basis_channel chn) {
   int len;
   char *r;
@@ -2020,3 +2120,22 @@ uw_Basis_string uw_Basis_makeSigString(uw_context ctx, uw_Basis_string sig) {
 uw_Basis_string uw_Basis_sigString(uw_context ctx, uw_unit u) {
   return uw_cookie_sig(ctx);
 }
+
+uw_Basis_string uw_Basis_fileName(uw_context ctx, uw_Basis_file f) {
+  return f.name;
+}
+
+uw_Basis_blob uw_Basis_fileData(uw_context ctx, uw_Basis_file f) {
+  return f.data;
+}
+
+uw_Basis_int uw_Basis_numFiles(uw_context ctx, uw_Basis_files fs) {
+  return fs.size;
+}
+
+uw_Basis_file uw_Basis_fileNum(uw_context ctx, uw_Basis_files fs, uw_Basis_int n) {
+  if (n < 0 || n >= fs.size)
+    uw_error(ctx, FATAL, "Files index out of bounds");
+  else
+    return fs.files[n];
+}