- [ ] `.FNT` bitmap fonts (`FONTS/*.FNT`)
- [ ] `.MIS` mission definitions (`M/*.MIS`)
- [ ] `.PIZ` hiscores (`D/HISCORES.PIZ`)
-- [ ] `T/*.E` text/dialog strings → structured extraction
+- [x] `T/*.E` text/dialog strings → `spec/formats/text-string-table.md`; all 28
+ extracted to `artifacts/text/*.json` (positional LF tables, cp437)
+- [ ] `Z/ARTIKEL*.E` newspaper articles (ISO-8859, `$` markup, leading code byte)
+ — different/richer format
+- [?] String index→in-game-use mapping (lives in `PT.EXE`; needs disassembly)
- [ ] Confirm whether `.PC`/`.BOB` reuse the VGA RLE codec
## Phase 2 — DOS environment spec (`spec/environment/`)
(not positional); the transparency/colorkey index; whether `.PC`/`.BOB` reuse
this RLE codec.
-**Artifacts:** `artifacts/*.pgm` (index-as-gray) and `*.ppm` (color). Tools:
-`tools/openpt/{rle,vga,palette,pnm,decode_vga,verify}.py`.
+**Artifacts:** `artifacts/vga/*.pgm` (index-as-gray). Tools:
+`tools/openpt/{rle,vga,palette,pnm,decode_vga,verify,build}.py`.
+
+**Tooling invocation settled:** run from repo root as
+`python3 -m tools.openpt.<cmd> PIZZA` (defaults `PIZZA`/`artifacts` relative to
+root). Getting-started in `tools/README.md`. (User rejected an auto path-resolution
+helper — keep invocation explicit/simple.)
+
+---
+
+## 2026-06-29 — Session 1 (cont.): text string tables
+
+**`T/*.E` string tables — DONE (high confidence):** `spec/formats/text-string-table.md`.
+- Positional table: LF(0x0A)-delimited, record N (0-based) = string index N.
+ Empty lines are **empty slots** (preserve indexing), not separators. Trailing
+ LF doesn't add a slot. No headers/markup; value substitution done in engine
+ code (no `%`-specifiers anywhere).
+- Extension = locale (`.E` = English). Files pure 7-bit ASCII; cp437 assumed for
+ localized builds (inferred).
+- Extracted all 28 tables to `artifacts/text/*.json` via `tools/openpt/text.py`
+ (also reports anomalies). `T/MAESTRO.E` flagged: one stray CR + one 0x81 byte
+ (harmless data glitch).
+- `.E` is overloaded: `GFX/PALETTE.E` is binary; `Z/ARTIKEL*.E` are ISO-8859
+ newspaper articles with `$` markup + leading code byte — a separate, richer
+ format, deferred (in CHECKLIST).
+- String index→in-game-use mapping is in `PT.EXE` (needs disassembly; out of scope).
--- /dev/null
+# Text string table (`T/*.E`)
+
+Positional tables of localized game strings: dialog, menu labels, messages.
+One file per subsystem (`BANK.E`, `WAFFEN.E`, `MARKT.E`, …), 28 files under `T/`.
+
+The file **extension is the locale**: `.E` = English. Other language builds use
+the corresponding suffix (cf. the in-binary templates `chr\c%de.pc`,
+`bob\typ%de.bob`, where the trailing `e` is the same English marker). *(inferred
+from naming + the engine's per-locale filename templates; confidence medium.)*
+
+## Layout *(provenance: observed; confidence: high)*
+
+Plain text. A file is a sequence of **records terminated by `LF` (0x0A)**:
+
+- Record *N* (0-based, counting from the start of the file) is **string index
+ *N***. The game looks strings up by this position.
+- An **empty record** (an `LF` with nothing before it) is an **empty string
+ slot** that preserves indexing — not a separator. (E.g. `T/PERS.E` indices
+ 12–17 are intentionally empty.)
+- Files end with a trailing `LF`; that final terminator does **not** add an
+ extra empty slot.
+- No length prefix, no count header, no in-text markup. Value substitution
+ (amounts, names) is performed by the engine concatenating around a string,
+ not via format codes in the text. *(observed: e.g. `BANK.E` "...fixed cost
+ is" with the number appended in code; no `%`-specifiers occur in any file.)*
+
+### Encoding
+
+DOS **code page 437**. The shipped English (`.E`) files are pure 7-bit ASCII (a
+subset of cp437), so this only matters for accented characters in localized
+builds. *(ASCII observed; cp437 inferred; confidence medium.)*
+
+## Algorithm (pseudocode)
+
+```
+parse(bytes) -> strings:
+ text = decode(bytes, codepage_437)
+ lines = text.split("\n")
+ if lines is non-empty and lines[-1] == "":
+ remove lines[-1] # the trailing-LF slot, not a real entry
+ return lines # strings[i] is string index i
+```
+
+## Test vector
+
+Input bytes `6f 66 20 35 30 2c 0a 6f 66 20 35 30 30 2c 0a`
+(`"of 50,\nof 500,\n"`) → `["of 50,", "of 500,"]`.
+
+Input `61 0a 0a 62 0a` (`"a\n\nb\n"`) → `["a", "", "b"]` (index 1 is an empty
+slot). These mirror real entries in `T/BANK.E` (indices 8–9) and the empty
+slots in `T/PERS.E`.
+
+## Anomalies / open questions
+
+- `T/MAESTRO.E` contains one stray `CR` (0x0D) and one non-ASCII byte (0x81) —
+ almost certainly data glitches in the original; parsing is unaffected.
+ *(observed.)*
+- The `.E` extension is **overloaded** and these are *not* the only `.E` files:
+ - `GFX/PALETTE.E` is binary (the palette bank — see [vga-image.md](vga-image.md)).
+ - `Z/ARTIKEL*.E` are ISO-8859 newspaper-article text with inline markup (a
+ leading code byte, `$` field markers) — a **different, richer format**, to
+ be documented separately. Tracked in [CHECKLIST](../../CHECKLIST.md).
+- Mapping of string **indices to their in-game uses** lives in `PT.EXE` and is
+ out of scope here (needs disassembly).
+
+## Reference tooling
+
+- `tools/openpt/text.py` — parser + extractor + anomaly scan; writes
+ `artifacts/text/<NAME>.json` (`{file, strings[]}`).
--- /dev/null
+"""Parser/extractor for Pizza Tycoon's string tables (T/*.E).
+
+A string table is a positional list of game strings: LF-delimited, one string
+per line, indexed by 0-based line number. Empty lines are empty string slots
+that preserve indexing. See spec/formats/text-string-table.md.
+
+The shipped English tables (the `E` suffix = English locale) are pure ASCII;
+text is decoded as DOS code page 437, which is a superset of ASCII and is the
+correct codepage for the localized (e.g. German) builds.
+
+Run from the repo root:
+ python3 -m tools.openpt.text [PIZZA_DIR] [ARTIFACTS_DIR]
+will scan every T/*.E, report anomalies, and write artifacts/text/<NAME>.json.
+"""
+
+import glob
+import json
+import os
+import sys
+
+ENCODING = "cp437"
+
+
+def parse(blob):
+ """Decode a string-table file into a list of strings (index = position)."""
+ text = blob.decode(ENCODING)
+ lines = text.split("\n")
+ if lines and lines[-1] == "":
+ lines.pop() # drop the slot created by the trailing newline
+ return lines
+
+
+def scan(pizza):
+ """Yield (path, strings, anomalies) for each T/*.E file."""
+ for path in sorted(glob.glob(os.path.join(pizza, "T", "*.E"))):
+ blob = open(path, "rb").read()
+ anomalies = []
+ if not blob.endswith(b"\n"):
+ anomalies.append("no trailing LF")
+ if b"\r" in blob:
+ anomalies.append("contains CR")
+ if any(b >= 0x80 for b in blob):
+ anomalies.append("non-ASCII byte(s)")
+ yield path, parse(blob), anomalies
+
+
+def main(argv):
+ pizza = argv[1] if len(argv) > 1 else "PIZZA"
+ artifacts = argv[2] if len(argv) > 2 else "artifacts"
+ out_dir = os.path.join(artifacts, "text")
+ os.makedirs(out_dir, exist_ok=True)
+ total = 0
+ for path, strings, anomalies in scan(pizza):
+ name = os.path.splitext(os.path.basename(path))[0]
+ with open(os.path.join(out_dir, name + ".json"), "w", encoding="utf-8") as f:
+ json.dump({"file": os.path.basename(path), "strings": strings},
+ f, ensure_ascii=False, indent=1)
+ total += 1
+ flag = (" <- " + ", ".join(anomalies)) if anomalies else ""
+ print("%-14s %4d strings%s" % (name, len(strings), flag))
+ print("text: wrote %d JSON tables to %s" % (total, out_dir))
+
+
+if __name__ == "__main__":
+ main(sys.argv)
projects / openpt.git / commitdiff