fix(rtl): les textes centrs en persan s'affichaient du mauvais cote

- Word: sous w:bidi, w:jc est logique (left=dbut=droite visuelle,
  right=fin=gauche visuelle) ; forcer jc=right alignait donc les
  paragraphes a gauche et crasait le centrage hrit du style.
  Dsormais le RTL pose w:bidi et w:rtl sans jamais toucher w:jc,
  comme le fait Word lui-mme pour un document RTL.
- PowerPoint: algn n'est plus crit quand le paragraphe n'en dfinit
  pas, afin de respecter l'alignement hrit du masque (titres centrs) ;
  algn=l explicite reste converti en r (algn est visuel en DrawingML).
- tests: centrage par style Word prserv, aucun w:jic crit, algn
  absent non cras côté PowerPoint

translators/pptx_translator.py

@@ -73,9 +73,15 @@ def _set_pptx_paragraph_rtl(paragraph) -> None:
     Enable RTL mode on a PowerPoint paragraph.
 
     Sets rtl="1" on the <a:pPr> element, which controls the paragraph
-    text direction. The horizontal alignment (algn) is only forced to
-    "r" when the paragraph has no explicit alignment or is aligned
-    "l" (left) — a centered or justified title keeps its alignment.
+    text direction; with rtl="1" and no explicit algn, PowerPoint
+    renders the paragraph from its start edge — the right edge.
+
+    algn is only rewritten when the paragraph explicitly says "l": in
+    DrawingML algn is visual (not logical like Word's w:jc), so an
+    explicit left would keep RTL text on the left edge. When algn is
+    absent, the inherited layout/master alignment (often centered
+    titles) must keep playing — writing "r" here would override it,
+    which is the centered-text mis-rendering reported for Persian.
     """
     p_elem = paragraph._p
     tag_pPr = f"{{{_NS_A}}}pPr"
@@ -84,7 +90,7 @@ def _set_pptx_paragraph_rtl(paragraph) -> None:
         pPr = etree.Element(tag_pPr)
         p_elem.insert(0, pPr)
     pPr.set("rtl", "1")
-    if pPr.get("algn") in (None, "", "l"):
+    if pPr.get("algn") == "l":
         pPr.set("algn", "r")
 
 

translators/word_translator.py

@@ -153,8 +153,6 @@ _PPR_BIDI_SUCCESSORS = (
     qn("w:sectPr"),
     qn("w:pPrChange"),
 )
-# Elements that must come AFTER w:jc inside w:pPr (w:jc itself excluded).
-_PPR_JC_SUCCESSORS = _PPR_BIDI_SUCCESSORS[_PPR_BIDI_SUCCESSORS.index(qn("w:jc")) + 1 :]
 # Elements that must come AFTER w:rtl inside w:rPr.
 _RPR_RTL_SUCCESSORS = (
     qn("w:cs"),
@@ -205,11 +203,18 @@ def _set_paragraph_rtl(paragraph: Paragraph) -> None:
 
     Sets:
       - w:pPr/w:bidi  → paragraph text direction = RTL
-      - w:pPr/w:jc    → mirrored alignment (left→right), ONLY when the
-        paragraph has no explicit alignment — centered/justified titles
-        must not be forced right-aligned.
       - w:rPr/w:rtl   → run-level RTL marker for each run
 
+    w:jc is deliberately left untouched. Under w:bidi, Word interprets
+    w:jc LOGICALLY (start/end), not visually: "left" means the start —
+    the RIGHT edge for an RTL paragraph — and "right" means the end, the
+    LEFT edge. Forcing jc="right" therefore aligns paragraphs to the
+    visual left and overrides style-level centering, which is exactly
+    the mis-rendering reported for centered Persian text. With bidi
+    alone: unaligned paragraphs start at the right edge, "left" resolves
+    to the right edge, and center/justify stay centered/justified — the
+    same markup Word itself writes for RTL documents.
+
     Every element is inserted at its schema position inside w:pPr/w:rPr
     (OOXML requires a fixed child order; appended-at-the-end elements
     make Word flag the file for repair).
@@ -219,14 +224,6 @@ def _set_paragraph_rtl(paragraph: Paragraph) -> None:
     if pPr.find(qn("w:bidi")) is None:
         _insert_ordered(pPr, OxmlElement("w:bidi"), _PPR_BIDI_SUCCESSORS)
 
-    jc = pPr.find(qn("w:jc"))
-    explicit_alignment = jc is not None and jc.get(qn("w:val")) not in (None, "", "left")
-    if not explicit_alignment:
-        if jc is None:
-            jc = OxmlElement("w:jc")
-            _insert_ordered(pPr, jc, _PPR_JC_SUCCESSORS)
-        jc.set(qn("w:val"), "right")
-
     for run in paragraph.runs:
         rPr = run._r.get_or_add_rPr()
         if rPr.find(qn("w:rtl")) is None: