"""05_make_tables.py -- emit LaTeX table fragments for the report."""
import json, pandas as pd, numpy as np
T="results/tables"; R="paper/tables"
RES=json.load(open(f"{T}/results.json"))

def w(name, body): open(f"{R}/{name}.tex","w").write(body)

# --- Table 1: summary stats ---
s=pd.read_csv(f"{T}/t1_summary_stats.csv",index_col=0)
nice={"ai_adoption_wmean":"AI adoption (wmean)","ai_infrastructure_wmean":"AI infrastructure (wmean)",
      "ai_disruption_risk_wmean":"AI disruption risk (wmean)","fwd_ret":"Forward 12m return",
      "fwd_ret_mktadj":"Forward return, mkt-adj.","mom_12m":"Momentum (12m)","vol_12m":"Volatility (ann.)",
      "log_mktcap":"Log market cap"}
rows="".join(f"{nice.get(i,i)} & {s.loc[i,'mean']:.3f} & {s.loc[i,'std']:.3f} & {s.loc[i,'5%']:.3f} & {s.loc[i,'50%']:.3f} & {s.loc[i,'95%']:.3f} \\\\\n" for i in nice)
w("tab_summary", r"""\begin{table}[H]\centering\small
\caption{Summary statistics, firm-year panel with tradeable forward returns (N=%d firm-years, %d firms, formation years %s).}\label{tab:summ}
\begin{tabular}{lrrrrr}\toprule
Variable & Mean & SD & p5 & Median & p95 \\\midrule
%s\bottomrule
\end{tabular}\end{table}""" % (RES["n_obs"],RES["n_firms"],RES["yrs"],rows))

# --- Table 2: portfolio sorts ---
sd=pd.read_csv(f"{T}/t2_portfolio_sorts.csv",index_col=0)
def row(key,label):
    r=sd.loc[key]
    return f"{label} & {r.Q1*100:.1f} & {r.Q2*100:.1f} & {r.Q3*100:.1f} & {r.Q4*100:.1f} & {r.Q5*100:.1f} & {r.Q5_minus_Q1*100:+.1f} & {r.t_stat:.2f} \\\\\n"
body =row("AI_ADOPT_EW","AI adoption (EW)")
body+=row("AI_ADOPT_VW","\\quad value-weighted")
body+=row("AI_INFRA_EW","AI infrastructure (EW)")
body+=row("AI_INFRA_VW","\\quad value-weighted")
body+=row("AI_RISK_EW","AI disruption risk (EW)")
body+=row("AI_RISK_VW","\\quad value-weighted")
body+="\\midrule\n"+row("AI_COMPOS_EW","AI composite (EW)")
body+=row("AI_COMPOS_VW","\\quad value-weighted")
w("tab_sorts", r"""\begin{table}[H]\centering\small
\caption{Forward 12-month returns (\%%) by AI-disclosure quintile. Q5$-$Q1 is the high-minus-low spread; \textit{t} uses Newey--West (3 lag) errors on the 15 annual spreads.}\label{tab:sorts}
\begin{tabular}{lrrrrrrr}\toprule
Signal (weighting) & Q1 & Q2 & Q3 & Q4 & Q5 & Q5$-$Q1 & \textit{t} \\\midrule
%s\bottomrule
\end{tabular}\end{table}""" % body)

# --- Table 3: factor alphas ---
a=RES["ls_alpha"]
def arow(k,label):
    x=a[k]
    return (f"{label} & {x['mean_ls_ann']*100:+.1f} & {x['capm_alpha_m']*1200:+.1f} & {x['capm_alpha_t']:.2f} "
            f"& {x['ff6_alpha_ann']*100:+.1f} & {x['ff6_alpha_t']:.2f} \\\\\n")
body=arow("AI_ADOPT","AI adoption")+arow("AI_INFRA","AI infrastructure")+arow("AI_COMPOS","AI composite")+arow("AI_RISK","AI disruption risk")
w("tab_alpha", r"""\begin{table}[H]\centering\small
\caption{Risk-adjusted performance of the equal-weight high-minus-low (Q5$-$Q1) long-short portfolio, annually rebalanced (%d monthly observations). Alphas annualized; \textit{t} from Newey--West (3 lag) HAC errors.}\label{tab:alpha}
\begin{tabular}{lrrrrr}\toprule
& Mean LS & CAPM $\alpha$ & \textit{t} & FF6 $\alpha$ & \textit{t} \\
Signal & (\%%/yr) & (\%%/yr) & & (\%%/yr) & \\\midrule
%s\bottomrule
\end{tabular}\end{table}""" % (a["AI_ADOPT"]["n_months"], body))

# --- Table 4: panel FE + Fama-MacBeth ---
fe=RES["panel_fe"]; fm=RES["fama_macbeth"]
def g(d,v):
    return d["coefs"].get(v)
m5=fe["M5_all"]; fmall=fm["all"]
def cell(coef,t): return f"{coef:+.4f} \\tstat{{{t:.2f}}}"
rows=[]
labels=[("z_AI_ADOPT","AI adoption ($z$)"),("z_AI_INFRA","AI infrastructure ($z$)"),
        ("z_AI_RISK","AI disruption risk ($z$)"),("z_size","Size ($z$)"),
        ("z_mom","Momentum ($z$)"),("z_vol","Volatility ($z$)")]
for v,lab in labels:
    fec=m5["coefs"].get(v); fmc=fmall.get(v)
    fecell = cell(fec["coef"],fec["t"]) if fec else "---"
    fmcell = cell(fmc["mean_coef"],fmc["t"]) if fmc else "---"
    rows.append(f"{lab} & {fecell} & {fmcell} \\\\")
body="\n".join(rows)
w("tab_panel", r"""\begin{table}[H]\centering\small
\caption{Forward-return regressions. Column (1): two-way (firm $\times$ year) fixed-effects panel, SE double-clustered by firm and year. Column (2): Fama--MacBeth, Newey--West \textit{t}. Dependent variable is the winsorized forward 12-month return; signals are within-year $z$-scores.}\label{tab:panel}
\begin{tabular}{lcc}\toprule
& (1) Two-way FE & (2) Fama--MacBeth \\\midrule
%s
\midrule
Firm FE & Yes & No \\
Year FE & Yes & (by construction) \\
N (obs / years) & %d & %d \\
\bottomrule
\end{tabular}\end{table}""" % (body, fe["M5_all"]["n"], fmall["n_years"]))

# --- Table 5: structural break ---
br=RES["structural_break"]; sp=RES["subperiod_spread"]
def brow(k,label):
    b=br[k]; s=sp[k]
    pre=s["pre_2010_2021"]; post=s["post_2022_2024"]
    return (f"{label} & {pre[0]*100:+.1f} \\tstat{{{pre[1]:.2f}}} & {post[0]*100:+.1f} \\tstat{{{post[1]:.2f}}} "
            f"& {b['interaction']:+.4f} \\tstat{{{b['interaction_t']:.2f}}} \\\\\n")
body=brow("AI_ADOPT","AI adoption")+brow("AI_INFRA","AI infrastructure")+brow("AI_COMPOS","AI composite")
w("tab_break", r"""\begin{table}[H]\centering\small
\caption{The post-ChatGPT structural break. Columns 2--3: Q5$-$Q1 forward-return spread (\%%/yr, Newey--West \textit{t}) in each sub-period. Column 4: coefficient on AI~$z$\,$\times$\,Post-2022 from a pooled OLS (size and momentum controls, year-clustered SE).}\label{tab:break}
\begin{tabular}{lccc}\toprule
& \multicolumn{2}{c}{Q5$-$Q1 spread} & Interaction \\
\cmidrule(lr){2-3}
Signal & Pre (2010--21) & Post (2022--24) & AI$\times$Post \\\midrule
%s\bottomrule
\end{tabular}\end{table}""" % body)

print("LaTeX table fragments written.")