feat: implement Phase 1 — Python/R parser, Julia codegen, and benchmarks

Implement the core julianiser pipeline that analyses Python/R data-science
code and generates equivalent Julia modules with type annotations:

- Manifest: [project], [[sources]], [[mappings]], [julia] sections
- Parser: detects pandas/numpy/scipy/dplyr/ggplot2/tidyr library calls
  from import statements, qualified calls, and bare function calls
- Julia codegen: generates .jl modules with using statements, translated
  function calls, Project.toml with package UUIDs
- Benchmarks: generates BenchmarkTools.jl script, shell runner comparing
  original vs Julia, and results.toml template
- ABI: SourceLanguage, LibraryMapping, JuliaType, TranslationUnit,
  BenchmarkResult, DetectedCall domain types
- Tests: 25 unit tests + 8 integration tests + 1 doctest (59 total)
- Example: data-pipeline with Python pandas/numpy and R dplyr/ggplot2

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: hyperpolymath

Cargo.lock

@@ -21,3 +21,4 @@ walkdir = "2"
 
 [dev-dependencies]
 tempfile = "3"
+serde_json = "1"

Cargo.toml

@@ -0,0 +1,38 @@
+# SPDX-License-Identifier: PMPL-1.0-or-later
+# Example R analysis script for julianiser translation demo.
+#
+# This script demonstrates common dplyr/ggplot2 patterns that julianiser
+# can detect and translate into Julia equivalents.
+
+library(dplyr)
+library(ggplot2)
+
+# Load and filter data.
+data <- read_csv("experiment_results.csv")
+clean_data <- filter(data, !is.na(value))
+
+# dplyr pipeline: group, summarise, arrange.
+summary_stats <- clean_data %>%
+  group_by(treatment) %>%
+  summarise(
+    mean_value = mean(value),
+    sd_value = sd(value),
+    n = n()
+  ) %>%
+  arrange(desc(mean_value))
+
+# Join with metadata.
+metadata <- read_csv("metadata.csv")
+enriched <- left_join(summary_stats, metadata, by = "treatment")
+
+# Statistical test.
+model <- lm(value ~ treatment + age + gender, data = clean_data)
+
+# Visualisation.
+p <- ggplot(clean_data, aes(x = treatment, y = value, fill = treatment)) +
+  geom_bar(stat = "summary", fun = "mean") +
+  geom_point(position = position_jitter(width = 0.2), alpha = 0.5) +
+  theme(legend.position = "none")
+
+write_csv(enriched, "enriched_results.csv")
+print(summary_stats)

examples/data-pipeline/analysis.R

@@ -0,0 +1,38 @@
+# SPDX-License-Identifier: PMPL-1.0-or-later
+# julianiser manifest — Example data pipeline translation
+#
+# This example demonstrates translating a Python pandas/numpy pipeline
+# and an R dplyr/ggplot2 analysis script into Julia equivalents.
+
+[project]
+name = "data-pipeline-example"
+version = "0.1.0"
+description = "Example: translate Python and R data pipelines to Julia"
+
+[[sources]]
+path = "pipeline.py"
+language = "python"
+
+[[sources]]
+path = "analysis.R"
+language = "r"
+
+[[mappings]]
+from-lib = "pandas"
+to-lib = "DataFrames.jl"
+
+[[mappings]]
+from-lib = "numpy"
+to-lib = "Base"
+
+[[mappings]]
+from-lib = "dplyr"
+to-lib = "DataFrames.jl"
+
+[[mappings]]
+from-lib = "ggplot2"
+to-lib = "Plots.jl"
+
+[julia]
+version = "1.10"
+packages = ["DataFrames", "CSV", "Statistics", "LinearAlgebra", "Plots", "BenchmarkTools"]

examples/data-pipeline/julianiser.toml

@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: PMPL-1.0-or-later
+# Example Python data pipeline for julianiser translation demo.
+#
+# This script demonstrates common pandas/numpy patterns that julianiser
+# can detect and translate into Julia equivalents.
+
+import pandas as pd
+import numpy as np
+
+# Load data from CSV.
+df = pd.read_csv("sales_data.csv")
+
+# Filter rows and compute derived columns.
+filtered = df.dropna()
+filtered["total"] = filtered["quantity"] * filtered["price"]
+
+# Group and aggregate.
+summary = filtered.groupby("category").agg({"total": "sum", "quantity": "mean"})
+
+# Numerical computation with numpy.
+prices = np.array(filtered["price"].values)
+quantities = np.array(filtered["quantity"].values)
+weighted_avg = np.mean(prices * quantities) / np.mean(quantities)
+
+# Statistical summary.
+std_dev = np.std(prices)
+correlation = np.dot(prices - np.mean(prices), quantities - np.mean(quantities))
+
+# Sort and output.
+result = summary.sort_values("total", ascending=False)
+result.to_csv("output.csv")
+
+print(f"Weighted average price: {weighted_avg:.2f}")
+print(f"Price std dev: {std_dev:.2f}")
+print(f"Price-quantity correlation: {correlation:.4f}")