vn-financial-terms

from vn_financial_terms import lookup, translate, search, by_domain, export

# ── Lookup by any surface form ──────────────────────────────────────────────
term = lookup("EBITDA")
print(term.en)       # Earnings before interest, taxes, depreciation, and amortization
print(term.vi)       # Lợi nhuận trước lãi vay, thuế, khấu hao và phân bổ
print(term.domain)   # accounting

# Works with Vietnamese abbreviations and alternate forms
lookup("TSCĐ").en          # "Fixed assets"
lookup("Tài sản lưu động") # resolves via alt_vi → Current assets
lookup("nợ xấu").en_abbr   # "NPL"

# ── Translate ───────────────────────────────────────────────────────────────
translate("Tài sản cố định", to="en")   # "Fixed assets"
translate("Revenue", to="vi")            # "Doanh thu"
translate("TNDN", to="en")             # "Corporate income tax"
translate("nonexistent", to="en")       # None

# ── Search (partial match) ───────────────────────────────────────────────────
results = search("thuế")                          # all terms containing "thuế"
results = search("income", domains=["tax"])        # domain-filtered
results = search("lợi nhuận", domains=["accounting", "banking"])

# ── Browse by domain ─────────────────────────────────────────────────────────
accounting_terms = by_domain("accounting")        # list[Term]
tax_terms        = by_domain("tax")

# ── Bulk export ──────────────────────────────────────────────────────────────
csv_str  = export("csv",      source="vi", target="en")
json_str = export("json",     source="vi", target="en", domains=["banking"])
md_str   = export("markdown", source="vi", target="en", domains=["tax"])

lookup("Revenue")               # by canonical EN
lookup("doanh thu")             # by canonical VI (case-insensitive)
lookup("EBITDA")                # by abbreviation
lookup("TSCĐ")                  # by VI abbreviation
lookup("Tài sản dài hạn hữu hình")  # by alt_vi
lookup("  Revenue  ")           # whitespace stripped
lookup("")                      # returns None
lookup("nonexistent")           # returns None

translate("Doanh thu", to="en")    # "Revenue"
translate("Revenue", to="vi")      # "Doanh thu"
translate("TNDN", to="en")        # "Corporate income tax"
translate("nope", to="en")         # None
translate("x", to="fr")            # raises ValueError

search("thuế")                         # all terms with "thuế" anywhere
search("income", domains=["tax"])      # narrow to one domain
search("")                             # returns []

by_domain("accounting")   # 55 terms
by_domain("BANKING")      # same as by_domain("banking")
by_domain("invalid")      # []

export("csv",      source="vi", target="en")
export("json",     source="en", target="vi", domains=["tax", "accounting"])
export("markdown", source="vi", target="en", domains=["banking"])

@dataclass(frozen=True)
class Term:
    en:            str               # canonical English term
    vi:            str               # canonical Vietnamese term
    domain:        str               # one of 8 allowed domains
    en_abbr:       str | None        # English abbreviation (e.g. "EBITDA")
    vi_abbr:       str | None        # Vietnamese abbreviation (e.g. "TNDN")
    alt_en:        tuple[str, ...]   # alternate English surface forms
    alt_vi:        tuple[str, ...]   # alternate Vietnamese surface forms
    vas_ref:       str | None        # VAS reference (e.g. "VAS 14")
    ifrs_ref:      str | None        # IFRS/IAS reference (e.g. "IFRS 15")
    definition_en: str | None        # plain-English definition
    definition_vi: str | None        # Vietnamese definition
    notes:         str | None        # translation guidance, register caveats

    # Convenience properties
    all_en_forms:  tuple[str, ...]   # en + en_abbr + alt_en
    all_vi_forms:  tuple[str, ...]   # vi + vi_abbr + alt_vi

    def to_dict(self) -> dict[str, Any]: ...  # None/empty fields omitted

from vn_financial_terms import TerminologyGuide

guide = TerminologyGuide(domains=["accounting", "tax", "banking"])
system_prompt = guide.build_system_prompt(source="vi", target="en")

import anthropic

client = anthropic.Anthropic()
response = client.messages.create(
    model="claude-opus-4-6",
    max_tokens=8096,
    system=system_prompt,          # <── glossary-anchored system prompt
    messages=[{"role": "user", "content": chapter_text}],
)
print(response.content[0].text)

from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": system_prompt},
        {"role": "user",   "content": chapter_text},
    ],
)

# Full book — all 230+ terms
full_guide = TerminologyGuide()

# Chapter 3 covers accounting and tax only
chapter_guide = TerminologyGuide(domains=["accounting", "tax"])

# Real-estate chapter
re_guide = TerminologyGuide(domains=["real_estate", "regulatory"])

# Or filter an existing guide
banking_guide = full_guide.filter(["banking"])

guide = TerminologyGuide(domains=["banking"])

# SDL Trados / memoQ / OmegaT
guide.to_csv(source="vi", target="en")

# JSON for vector databases / custom pipelines
guide.to_json(source="vi", target="en")

# Markdown table for Notion docs or inline prompt injection
guide.to_markdown(source="vi", target="en")

block = guide.build_glossary_block(source="vi", target="en", fmt="markdown")
user_message = f"""Translate the following using this glossary:

{block}

---

{chapter_text}
"""