4 files changed, 83 insertions(+), 13 deletions(-)

M rhapsode.cabal
M src/CExports.hs
M src/Links.hs
M src/main.c

@@ 65,7 65,7 @@ library
         network-uri,
         stylist >= 2.4 && <3, css-syntax, xml-conduit-stylist >= 2.3 && <3, scientific,
         async, hurl >= 2, filepath, temporary,
-        file-embed >= 0.0.9 && < 0.1, time, parallel >= 1
+        file-embed >= 0.0.9 && < 0.1, time, parallel >= 1, process
   
   -- Directories containing source files.
   hs-source-dirs:      src

@@ 38,10 38,10 @@ import Foreign.Marshal.Array
 --pair a b = (a, b)
 
 -- FIXME: Segfaults, was intended for the sake of easy concurrency.
-foreign export ccall c_docLinksAndRendering :: StablePtr Session -> StablePtr Page -> Bool -> IO (CArray CString)
+foreign export ccall c_docLinksAndRendering :: StablePtr Session -> StablePtr Page -> Bool -> CString -> IO (CArray CString)
 
-c_docLinksAndRendering c_session c_page rewriteUrls = do
-    c_links <- c_extractLinks c_page
+c_docLinksAndRendering c_session c_page rewriteUrls c_v2jProfile = do
+    c_links <- c_extractLinks c_page c_v2jProfile
     ssml <- c_renderDoc c_session c_page rewriteUrls
     -- (c_links, ssml) <- c_extractLinks c_page `concurrently` c_renderDoc c_session c_page rewriteUrls
     nil <- newCString ""

@@ 29,6 29,10 @@ import qualified Data.Set as Set
 import Data.List (nub, intercalate)
 import Control.Concurrent (forkIO)
 
+-- For Voice2Json's sentences.ini
+import Data.Char
+import System.Process (callProcess)
+
 data Link = Link {
     label :: Text,
     title :: Text,


@@ 208,14 212,34 @@ updateSuggestions page = do
     createDirectoryIfMissing True dir
     Prelude.writeFile path $ unlines $ map unwords suggestions'
 
+-------
+--- Voice2Json language models
+-------
+
+-- | Output links to a Voice2Json sentences.ini grammar.
+outputSentences _ "" = return ()
+outputSentences links dir = do
+        Prelude.writeFile (dir </> "sentences.ini") $ unlines sentences
+        callProcess "voice2json" ["voice2json", "--profile", dir, "train-profile"]
+    where
+      sentences = "[links]" : [
+            unwords $ words $ filter validChar line -- Enforce valid sentences.ini syntax.
+            | line@(_:_) <- map (unpack . label) links ++ map (unpack . title) links ++ map (show . href) links
+        ]
+      -- | Can this character appear in a sentences.ini rule without escaping?
+      validChar ch = not (isAscii ch) || isSpace ch || isAlphaNum ch
+
 -- C API
-foreign export ccall c_extractLinks :: StablePtr Page -> IO (CArray CString)
+foreign export ccall c_extractLinks :: StablePtr Page -> CString -> IO (CArray CString)
 
-c_extractLinks c_page = do
+c_extractLinks c_page c_v2jProfile = do
     page <- deRefStablePtr c_page
+    v2jProfile <- peekCString c_v2jProfile
     forkIO $ updateSuggestions page -- background process for useful navigation aid.
     bookmarks <- readBookmarks
-    ret <- forM (linksFromPage page ++ extractLinks bookmarks) $ \link -> do
+    let links = linksFromPage page ++ extractLinks bookmarks
+    forkIO $ outputSentences links v2jProfile
+    ret <- forM links $ \link -> do
         c_label <- text2cstring $ strip $ label link
         c_title <- text2cstring $ strip $ title link
         c_href <- newCString $ uriToString id (href link) ""


@@ 248,3 272,11 @@ c_formatLink c_label c_title c_url = do
     audio src = el "audio" [("src", pack src)] []
     prosody attrs txt = el "prosody" attrs [NodeContent txt]
     style field mode inner = el "tts:style" [("field", field), ("mode", mode)] [NodeElement inner]
+
+foreign export ccall c_dataDir :: CString -> IO CString
+
+-- | Used to find Voice2Json profile
+c_dataDir c_subdir = do
+    subdir <- peekCString c_subdir
+    cache <- getXdgDirectory XdgData "rhapsode"
+    newCString (cache </> subdir)

@@ 6,6 6,9 @@
 #include <termios.h>
 #include <limits.h>
 
+#include <dirent.h>
+#include <errno.h>
+
 #include "HsFFI.h"
 #include <espeak-ng/espeak_ng.h>
 #include <sndfile.h>


@@ 31,10 34,11 @@ struct session *c_enableLogging(struct session*);
 void c_writeLog(char*, struct session*);
 
 char *c_renderDoc(struct session*, struct page*, _Bool);
-char **c_extractLinks(struct page*);
+char **c_extractLinks(struct page*, char *v2jProfile);
 char **c_docLinksAndRendering(struct session*, struct page*, _Bool); // FIXME segfaults.
 int c_ssmlHasMark(char*, char*);
 char *c_formatLink(char *label, char *title, char *url);
+char *c_dataDir(char *subdir);
 
 char *c_lastVisited(char*);
 


@@ 322,16 326,25 @@ FILE *parse_opt_file() {
     return ret;
 }
 
+int dir_exists(char *path) {
+    DIR *dir = opendir(path);
+    if (dir) closedir(dir);
+    else return ENOENT == errno;
+    return 1;
+}
+
 int main(int argc, char **argv) {
     int speak_err = 0;
     struct session *session = NULL;
     struct page *referer = NULL;
 
     int use_espeak = 0;
+    int validate_v2j_profile = 0;
     char *logpath = NULL;
     char *mimes;
     FILE *fd_ssml = NULL;
     FILE *fd_links = NULL;
+    char *v2j_profile = c_dataDir("voice2json");
 
     hs_init(&argc, &argv);
 


@@ 344,9 357,9 @@ int main(int argc, char **argv) {
     int c;
     opterr = 0;
     #ifdef WITH_SPEECHD
-    while ((c = getopt(argc, argv, "xs::l::L:kKw::dh")) != -1) {
+    while ((c = getopt(argc, argv, "xs::l::L:kKvVw::dh")) != -1) {
     #else
-    while ((c = getopt(argc, argv, "xs::l::kKw::h")) != -1) {
+    while ((c = getopt(argc, argv, "xs::l::kKv::Vw::h")) != -1) {
     #endif
         switch (c) {
         case 'x':


@@ 367,6 380,14 @@ int main(int argc, char **argv) {
         case 'k':
             read_keyboard = 1;
             break;
+        case 'V':
+            v2j_profile = "";
+            validate_v2j_profile = 0;
+            break;
+        case 'v':
+            if (optarg != NULL) v2j_profile = optarg;
+            validate_v2j_profile = 1;
+            break;
         case 'w':
             use_espeak = 1;
             path_wav = optarg;


@@ 387,8 408,10 @@ int main(int argc, char **argv) {
             fprintf(stderr, "\t\t\thttps://xkcd.com/1692/\n");
             fprintf(stderr, "\t-l\tlinks\tWrite extracted links to specifed file or stdout as TSV.\n");
             fprintf(stderr, "\t-L\tlog\tWrite (append) network request timing to specified filepath.\n");
-            fprintf(stderr, "\t-k\tkeyboard\tRead arrow key navigation & links from stdin. Default behaviour, noop.\n");
-            fprintf(stderr, "\t-K\tDon't read input from stdin.");
+            fprintf(stderr, "\t-k\tkeyboard in\tRead arrow key navigation & links from stdin. Default behaviour, noop.\n");
+            fprintf(stderr, "\t-K\t\tDon't read input from stdin.");
+            fprintf(stderr, "\t-v\tvoice in\tEnsure voice input is enabled & optionally sets the Voice2JSON profile.\n");
+            fprintf(stderr, "\t-V\t\tDon't listen for voice input.\n");
             fprintf(stderr, "\t-w\t.wav\tWrite an audio recording of the webpage, or (DEFAULT) immediately output through speakers.\n");
             #ifdef WITH_SPEECHD
             fprintf(stderr, "\t-d\tSpeechD\tSchedule page read via the SpeechD daemon. (BROKEN)\n");


@@ 431,6 454,21 @@ int main(int argc, char **argv) {
     referer = c_initialReferer();
 
     if (use_espeak) speak_err = speak_initialize();
+
+    struct stat sb;
+    if (validate_v2j_profile && dir_exists(v2j_profile)) {
+        // TODO Would packagers want a way to configure autodownloading of these profiles?
+        fprintf(stderr, "Voice2JSON profile is uninitialized! Voice recognition is now disabled.\n");
+        fprintf(stderr, "Please download a profile from: http://voice2json.org/#supported-languages\n");
+        fprintf(stderr, "And save the download in: %s\n", v2j_profile);
+        if (use_espeak) {
+            speak_text("Voice2JSON profile is uninitialized! Voice recognition is now disabled.",
+                espeakRATE, 0);
+        }
+
+        v2j_profile = "";
+    }
+
     char *ssml, **links, *uri;
     int read_links = 0;
     if (optind >= argc) {


@@ 451,8 489,8 @@ read_uri:
         if (logpath != NULL) session = c_enableLogging(session);
 
         struct page *page = c_fetchURL(session, mimes, referer, uri);
+        links = c_extractLinks(page, v2j_profile);
         ssml = c_renderDoc(session, page, use_espeak);
-        links = c_extractLinks(page);
 
         if (logpath != NULL) c_writeLog(logpath, session);