some bugfixes around gluing chunks

smucclaw · Oct 20, 2021 · b3b4b4f · b3b4b4f
1 parent 381de0f
commit b3b4b4f
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 13 deletions.
diff --git a/mengwong/mp/src/Lib.hs b/mengwong/mp/src/Lib.hs
@@ -19,7 +19,7 @@ import qualified Data.Vector as V
 import Generic.Data (Generic)
 import Data.Vector ((!), (!?))
 import Data.Maybe (fromMaybe, catMaybes)
-import Text.Pretty.Simple (pPrint)
+import Text.Pretty.Simple (pPrint, pShow)
 import Control.Monad (guard, when, forM_)
 import qualified AnyAll as AA
 import qualified Text.PrettyPrint.Boxes as Box
@@ -151,7 +151,7 @@ getStanzas esa = do
   rs <- esa
   let chunks = getChunks $ Location rs (0,0) ((0,0),(V.length (rs ! (V.length rs - 1)) - 1, V.length rs - 1))
       toreturn = extractRange <$> glueChunks chunks
-  -- traceM ("getStanzas: extracted range " ++ show toreturn)
+  traceM ("getStanzas: extracted range " ++ (Text.unpack $ pShow toreturn))
   return toreturn
 
 -- because sometimes a chunk followed by another chunk is really part of the same chunk.
@@ -176,9 +176,9 @@ getChunks loc@(Location rs (_cx,_cy) ((_lx,_ly),(_rx,ry))) =
                      , any (\row ->
                                V.any (\w -> w `elem` Text.words "EVERY PARTY MUST MAY WHEN INCLUDES MEANS IS IF UNLESS")
                                (rs ! row)
-                               ||
-                               V.all Text.null (rs ! row)
-                           )
+                           ) rows
+                       ||
+                       all (\row -> V.all Text.null (rs ! row))
                        rows
                      ]
       toreturn = setRange loc <$> wantedChunks
@@ -322,6 +322,7 @@ pRule = withDepth 1 $ do
   _ <- optional dnl
   try (pRegRule <?> "regulative rule")
     <|> (pConstitutiveRule <?> "constitutive rule")
+    <|> (eof >> return [])
 
 pConstitutiveRule :: Parser [Rule]
 pConstitutiveRule = debugName "pConstitutiveRule" $ do

diff --git a/mengwong/mp/test/README.org b/mengwong/mp/test/README.org
@@ -390,14 +390,18 @@ do the individual components work?
 :TABLE_EXPORT_FORMAT: orgtbl-to-csv
 :END:
 
-| EVERY | person          | ("singer") |   |
-| WHO   | walks           |            |   |
-| AND   | eats            |            |   |
-| MUST  |                 |            |   |
-|       |                 |            |   |
-|       |                 |            |   |
-| IF    | the King wishes |            |   |
-| ->    | sing            |            |   |
+|                                                     |                 |            |   |
+| A paragraph with irrelevant text should be ignored. |                 | Like so.   |   |
+|                                                     |                 |            |   |
+|                                                     |                 |            |   |
+| EVERY                                               | person          | ("singer") |   |
+| WHO                                                 | walks           |            |   |
+| AND                                                 | eats            |            |   |
+| MUST                                                |                 |            |   |
+|                                                     |                 |            |   |
+|                                                     |                 |            |   |
+| IF                                                  | the King wishes |            |   |
+| ->                                                  | sing            |            |   |
 
 ** legislative source citations
 :PROPERTIES:

diff --git a/mengwong/mp/test/blank-lines.csv b/mengwong/mp/test/blank-lines.csv
@@ -1,3 +1,7 @@
+,,,
+A paragraph with irrelevant text should be ignored.,,Like so.,
+,,,
+,,,
 EVERY,person,"(""singer"")",
 WHO,walks,,
 AND,eats,,