Indic all: update 4.3 final-Reph-reordering instructions to be consis…

…tently worded and reflect the seven-step algorithm as defined in the MS script-shaping docs. This commit regularized all step-e language -- about moving Reph to just before the first post-base matra or SM or Vedic sign after the Reph-position tag. Full algorithm is listed in Indic General. Do note, however, that as of this writing, there seem to be some differences between the MS docs algorithm and what HarfBuzz does on REPH_POS_AFTER_SUBJOINED scripts, step d, and what it does for all scripts on step e. See issue n8willis#48 in this repo for discussion.
alolita · Mar 27, 2020 · af71a0d · af71a0d
1 parent a6022f3
commit af71a0d
Show file tree

Hide file tree

Showing 11 changed files with 175 additions and 48 deletions.
diff --git a/opentype-shaping-bengali.md b/opentype-shaping-bengali.md
@@ -1283,8 +1283,15 @@ The algorithm for finding the final "Reph" position is
   - If no such non-ligated post-base consonant is found in the
     previous step, move the "Reph" to the position immediately before
     the first post-base matra, syllable modifier, or Vedic sign that
-    has a positioning tag of `POS_BEFORE_POST` or later. This
-    will be the final "Reph" position.
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Bengali incorporates the
+    > `REPH_POS_AFTER_SUBJOINED` shaping characteristic, this means
+    > any positioning tag of `POS_BEFORE_POST` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_BEFORE_POST`.
   - If no other location has been located in the previous steps, move
     the "Reph" to the end of the syllable.
 

diff --git a/opentype-shaping-devanagari.md b/opentype-shaping-devanagari.md
@@ -1239,8 +1239,15 @@ The algorithm for finding the final "Reph" position is
   - If no such non-ligated post-base consonant is found in the
     previous step, move the "Reph" to the position immediately before
     the first post-base matra, syllable modifier, or Vedic sign that
-    has a positioning tag of `POS_POSTBASE_CONSONANT` or later. This
-    will be the final "Reph" position.
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Devanagari incorporates the
+    > `REPH_POS_BEFORE_POST` shaping characteristic, this means
+    > any positioning tag of `POS_POSTBASE_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_POSTBASE_CONSONANT`.
   - If no other location has been located in the previous steps, move
     the "Reph" to the end of the syllable.
 

diff --git a/opentype-shaping-gujarati.md b/opentype-shaping-gujarati.md
@@ -1194,9 +1194,57 @@ shaping characteristic, this final position is immediately before any
 independent post-base consonant forms (meaning the first post-base
 consonant that has not formed a ligature with the syllable base).
 
-  - If the syllable does not have any post-base consonants, then the
-    final "Reph" position is immediately before the first post-base
-    matra, syllable modifier, or Vedic sign.
+The algorithm for finding the final "Reph" position is
+
+<!---
+
+  - Find the first explicit "Halant" between the first post-Reph
+    consonant and the last main consonant. Move the "Reph to the
+    position immediately after this "Halant".
+	- If a zero-width joiner (ZWJ) or a zero-width non-joiner (ZWNJ)
+      follows this "Halant", move the "Reph" to the position
+      immediately after the ZWJ or ZWNJ.
+--->
+
+  - Starting at the first post-"Reph" consonant, search forward looking
+    for the first explicit "Halant", ending the search when the base
+    consonant is encountered. If such an explicit "Halant" is found,
+    move the "Reph" to the position immediately after this
+    "Halant".
+	  * If a zero-width joiner (ZWJ) or a zero-width non-joiner (ZWNJ)
+        follows this "Halant", move the "Reph" to the position
+        immediately after the ZWJ or ZWNJ. This will be the final
+        "Reph" position. 
+	  * If no ZWJ or ZWNJ follows this "Halant", leave the "Reph" in
+        its position immediately after the "Halant". This will be the
+        final "Reph" position. 
+  - If no such explicit "Halant" is found in the previous step, find
+    the first post-base consonant that has not formed a ligature with
+    the base consonant. If such a non-ligated post-base consonant is
+    found, move the "Reph" to the position immediately before the
+    non-ligated post-base consonant. This will be the final "Reph"
+    position.
+  - If no such non-ligated post-base consonant is found in the
+    previous step, move the "Reph" to the position immediately before
+    the first post-base matra, syllable modifier, or Vedic sign that
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Gujarati incorporates the
+    > `REPH_POS_BEFORE_POST` shaping characteristic, this means
+    > any positioning tag of `POS_POSTBASE_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_POSTBASE_CONSONANT`.
+  - If no other location has been located in the previous steps, move
+    the "Reph" to the end of the syllable.
+
+
+Finally, if the final position of "Reph" occurs after a
+"_matra_,Halant" subsequence, then "Reph" must be repositioned to the
+left of "Halant", to allow for potential matching with `abvs` or
+`psts` substitutions from GSUB.
+
 
 
 

diff --git a/opentype-shaping-gurmukhi.md b/opentype-shaping-gurmukhi.md
@@ -1282,8 +1282,15 @@ The algorithm for finding the final "Reph" position is
   - If no such non-ligated post-base consonant is found in the
     previous step, move the "Reph" to the position immediately before
     the first post-base matra, syllable modifier, or Vedic sign that
-    has a positioning tag of `POS_BELOWBASE_CONSONANT` or later. This
-    will be the final "Reph" position.
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Gurmukhi incorporates the
+    > `REPH_POS_BEFORE_SUBJOINED` shaping characteristic, this means
+    > any positioning tag of `POS_BELOWBASE_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_BELOWBASE_CONSONANT`.
   - If no other location has been located in the previous steps, move
     the "Reph" to the end of the syllable.
 

diff --git a/opentype-shaping-indic-general.md b/opentype-shaping-indic-general.md
@@ -1323,7 +1323,7 @@ Reph-position shaping characteristic, and is conditional upon the
 presence or absence of certain characters (such as post-base
 consonants or "matra,Halant" sequences) in the syllable. 
 
-The algorithm for determining the final Reph position has seven steps.
+The full algorithm for determining the final Reph position has seven steps.
 
 (a) If the script uses Reph-position rule `REPH_POS_AFTER_POST`, jump
 immediately to step (e). Otherwise, proceed to step (b).
@@ -1353,8 +1353,10 @@ Reph-position rule, proceed to step (e).
 
 (e) Move the "Reph" to a position immediately before the first
 post-base matra, syllable modifier sign or Vedic sign that has a
-reordering class after the intended Reph position, then proceed to
-step (mH). If no such matra or sign is found, proceed to step (f).
+reordering class after the intended Reph position in the syllable sort
+order (as listed in [stage 2](#2-initial-reordering)). This will be
+the final "Reph" position. , then proceed to step (mH). If no such
+matra or sign is found, proceed to step (f).
 
 (f) Move the "Reph" to the end of the syllable. 
 

diff --git a/opentype-shaping-kannada.md b/opentype-shaping-kannada.md
@@ -1210,8 +1210,15 @@ The algorithm for finding the final "Reph" position is
 
   - Move the "Reph" to the position immediately before
     the first post-base matra, syllable modifier, or Vedic sign that
-    has a positioning tag of `POS_FINAL_CONSONANT` or later. This
-    will be the final "Reph" position.
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Kannada incorporates the
+    > `REPH_POS_AFTER_POST` shaping characteristic, this means
+    > any positioning tag of `POS_FINAL_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_FINAL_CONSONANT`.
   - If no other location has been located in the previous step, move
     the "Reph" to the end of the syllable.
 

diff --git a/opentype-shaping-malayalam.md b/opentype-shaping-malayalam.md
@@ -1265,11 +1265,17 @@ after the syllable base.
 
 The algorithm for finding the final "Reph" position is
 
-  - If no such non-ligated post-base consonant is found in the
-    previous step, move the "Reph" to the position immediately before
+  - Move the "Reph" to the position immediately before
     the first post-base matra, syllable modifier, or Vedic sign that
-    has a positioning tag of `POS_ABOVEBASE_CONSONANT` or later. This
-    will be the final "Reph" position.
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Malayalam incorporates the
+    > `REPH_POS_AFTER_MAIN` shaping characteristic, this means
+    > any positioning tag of `POS_ABOVEBASE_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_ABOVEBASE_CONSONANT`.
   - If no other location has been located in the previous step, move
     the "Reph" to the end of the syllable.
 

diff --git a/opentype-shaping-oriya.md b/opentype-shaping-oriya.md
@@ -1256,18 +1256,24 @@ position. Because Oriya incorporates the `REPH_POS_AFTER_MAIN`
 shaping characteristic, this final position is immediately after the
 syllable base.
 
-  - If the syllable does not have a base consonant (such as a syllable
-    based on an independent vowel), then the final "Reph" position is
-    immediately before the first character tagged with the
-    `POS_BEFORE_POST` position or any later position in the sort
-    order.
-
-    -- If there are no characters tagged with `POS_BEFORE_POST` or
-       later positions, then "Reph" is positioned at the end of the
-       syllable.
-
-Finally, if the final position of "Reph" occurs after a
-"_matra_,Halant" subsequence, then "Reph" must be repositioned to the
+The algorithm for finding the final "Reph" position is
+
+  - Move the "Reph" to the position immediately before
+    the first post-base matra, syllable modifier, or Vedic sign that
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Oriya incorporates the
+    > `REPH_POS_AFTER_MAIN` shaping characteristic, this means
+    > any positioning tag of `POS_ABOVEBASE_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_ABOVEBASE_CONSONANT`.
+  - If no other location has been located in the previous step, move
+    the "Reph" to the end of the syllable.
+
+Finally, if the final position of "Reph" or "Repha" occurs after a
+"_matra_,Halant" subsequence, then "Reph"/"Repha" must be repositioned to the
 left of "Halant", to allow for potential matching with `abvs` or
 `psts` substitutions from GSUB.
 

diff --git a/opentype-shaping-sinhala.md b/opentype-shaping-sinhala.md
@@ -1067,6 +1067,27 @@ position. Because Sinhala incorporates the `REPH_POS_AFTER_MAIN`
 shaping characteristic, this final position is immediately after the
 syllable base.
 
+The algorithm for finding the final "Reph" position is
+
+  - Move the "Reph" to the position immediately before
+    the first post-base matra, syllable modifier, or Vedic sign that
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Sinhala incorporates the
+    > `REPH_POS_AFTER_MAIN` shaping characteristic, this means
+    > any positioning tag of `POS_ABOVEBASE_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_ABOVEBASE_CONSONANT`.
+  - If no other location has been located in the previous step, move
+    the "Reph" to the end of the syllable.
+
+Finally, if the final position of "Reph" or "Repha" occurs after a
+"_matra_,Halant" subsequence, then "Reph"/"Repha" must be repositioned to the
+left of "Halant", to allow for potential matching with `abvs` or
+`psts` substitutions from GSUB.
+
 
 ![Reph positioning](/images/sinhala/sinhala-reph-position.png)
 

diff --git a/opentype-shaping-tamil.md b/opentype-shaping-tamil.md
@@ -1195,15 +1195,23 @@ position. Because Tamil incorporates the `REPH_POS_AFTER_POST`
 shaping characteristic, this final position is immediately after
 any post-base consonant forms.
 
-  - If the syllable does not have a base consonant (such as a syllable
-    based on an independent vowel), then the final "Reph" position is
-    immediately before the first character tagged with the
-    `POS_BEFORE_POST` position or any later position in the sort
-    order.
-
-    -- If there are no characters tagged with `POS_BEFORE_POST` or
-       later positions, then "Reph" is positioned at the end of the
-       syllable.
+
+The algorithm for finding the final "Reph" position is
+
+  - Move the "Reph" to the position immediately before
+    the first post-base matra, syllable modifier, or Vedic sign that
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Tamil incorporates the
+    > `REPH_POS_AFTER_POST` shaping characteristic, this means
+    > any positioning tag of `POS_FINAL_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_FINAL_CONSONANT`.
+  - If no other location has been located in the previous step, move
+    the "Reph" to the end of the syllable.
+
 
 Finally, if the final position of "Reph" occurs after a
 "_matra_,Halant" subsequence, then "Reph" must be repositioned to the

diff --git a/opentype-shaping-telugu.md b/opentype-shaping-telugu.md
@@ -1184,15 +1184,23 @@ position. Because Telugu incorporates the `REPH_POS_AFTER_POST`
 shaping characteristic, this final position is immediately after 
 any post-base consonant forms.
 
-  - If the syllable does not have a base consonant (such as a syllable
-    based on an independent vowel), then the final "Reph" position is
-    immediately before the first character tagged with the
-    `POS_BEFORE_POST` position or any later position in the sort
-    order.
-
-    -- If there are no characters tagged with `POS_BEFORE_POST` or
-       later positions, then "Reph" is positioned at the end of the
-       syllable.
+
+The algorithm for finding the final "Reph" position is
+
+  - Move the "Reph" to the position immediately before
+    the first post-base matra, syllable modifier, or Vedic sign that
+    has a positioning tag after the script's "Reph" position in the
+    syllable sort order (as listed in [stage
+    2](#2-initial-reordering)). This will be the final "Reph"
+    position. 
+	> Note: Because Telugu incorporates the
+    > `REPH_POS_AFTER_POST` shaping characteristic, this means
+    > any positioning tag of `POS_FINAL_CONSONANT` or later,
+    > although a post-base matra, syllable modifier, or Vedic sign
+    > would not typically be tagged with `POS_FINAL_CONSONANT`.
+  - If no other location has been located in the previous step, move
+    the "Reph" to the end of the syllable.
+
 
 Finally, if the final position of "Reph" occurs after a
 "_matra_,Halant" subsequence, then "Reph" must be repositioned to the